2018 Ltcmas, the Outlook for Global Synchronized Growth in More Than a Decade

FOR INSTITUTIONAL/WHOLESALE/PROFESSIONAL CLIENTS AND QUALIFIED INVESTORS ONLY – NOT FOR RETAIL USE OR DISTRIBUTION PORTFOLIO INSIGHTS

2018 | 22nd ANNUAL EDITION LONG-TERM CAPITAL MARKET ASSUMPTIONS Time-tested projections to build stronger portfolios

2018 LONG-TERM CAPITAL MARKET ASSUMPTIONS TABLE OF CONTENTS

4 FOREWORD EXECUTIVE SUMMARY 5 2018 Long-Term Capital Market Assumptions MACROECONOMIC ASSUMPTIONS 12 Mostly stable, mostly moderate

I. THEMATIC ARTICLES TECHNOLOGY, PRODUCTIVITY AND THE LABOR FORCE 18 The impact of technology on long-term potential economic growth THE IMPACT OF GLOBAL AGING 28 How demographic change will affect savings, growth and interest rates PENSION INVESTMENT STRATEGY 34 Matching cash flows and managing liquidity in maturing pension funds THE FUTURE PATH OF CHINESE INTEREST RATES 40 The cost of capital in China’s changing markets U.S. DOLLAR FORECAST 46 The path of the U.S. dollar: Looking forward by looking back

II. ASSUMPTIONS FIXED INCOME ASSUMPTIONS 54 G4 government bonds: Slow road, low yields EQUITY MARKET ASSUMPTIONS 60 Modestly but steadily lower returns ALTERNATIVE STRATEGY ASSUMPTIONS 65 A stable to improving alpha outlook offsets the beta drag CURRENCY EXCHANGE RATE ASSUMPTIONS 81 The tide has turned for the U.S. dollar VOLATILITY ASSUMPTIONS 86 Volatility: Cyclically lower, structurally unchanged

III. ASSUMPTIONS MATRICES 92 U.S. DOLLAR 94 STERLING 96 EURO

IV. APPENDIX 100 GLOSSARY 102 ACKNOWLEDGMENTS

J.P. MORGAN ASSET MANAGEMENT 3 FOREWORD

“PREDICTION IS VERY DIFFICULT,” Nobel Prize-winning physicist Niels Bohr wrote, “especially if it’s about the future.” Investors trying to assess the prospects for capital markets may sometimes share the sentiment.

Amid today’s challenging investing environment, we present the 2018 edition of J.P. Morgan Asset CHRIS WILLCOX Management’s Long-Term Capital Market Assumptions (LTCMAs). In our 22nd year of publishing capital market estimates, we incorporate more than 50 asset and strategy classes; our return assumptions are available in 13 base currencies, which this year include the Chinese renminbi for the first time. Over the years, many investors and advisors have come to depend on our assumptions to inform their strategic asset allocation, build stronger portfolios and establish reasonable expectations for risks and returns over a 10- to 15-year time frame.

We formulate our LTCMAs as part of a deeply researched proprietary process that draws on quantitative and qualitative inputs as well as insights from experts across J.P. Morgan Asset Management —a collaborative MIKE O’BRIEN effort that has been honed over the past two decades. Our own multi-asset investment approach relies heavily on our LTCMAs. The assumptions form a critical foundation of our framework for designing, building and analyzing solutions aligned with our clients’ specific investment needs.

In this edition of our assumptions, we explore the complex interplay between secular themes, including global aging and technological innovation, and cyclical factors—notably the slow path of policy normalization and elevated equity valuations—that will influence asset returns over our 10- to 15-year investment horizon. During this period, we see more modest returns in many asset markets. All of this highlights the importance of a considered, long-term strategic perspective. It also shines a light on the power of active asset allocation, capturing alternative sources of risk premium, and careful manager selection as investors look for new sources of potential returns.

We look forward to working with you to make the best use of our assumptions in setting your own strategic perspective and pursuing your investment goals.

On behalf of J.P. Morgan Asset Management, thank you for your continued trust and confidence. As always, we welcome your feedback.

Chris Willcox Mike O’Brien Chief Executive Officer, Chief Executive Officer, Asset Management Asset Management Solutions

4 LONG-TERM CAPITAL MARKET ASSUMPTIONS EXECUTIVE SUMMARY

2018 Long-Term Capital Market Assumptions John Bilton, CFA, Head of Global Multi-Asset Strategy, Multi-Asset Solutions

IN BRIEF This executive summary gives readers a broad overview of our 2018 Long-Term Capital Market Assumptions (LTCMAs) and provides a context for how some of the structural factors affecting economies today are likely to drive asset returns over a 10- to 15-year investment horizon. The key takeaways from this year’s LTCMAs:

• Our 2018 trend real GDP growth estimates of 1.5% in developed markets and 4.5% in emerging markets are unchanged from last year. However, beneath this stable outlook is evidence that the prolonged series of downgrades to trend growth, reflecting population aging, may be nearing an end. We also see potential for a technology-driven boost to productivity, which creates an upside risk to our forecasts. Yet despite cautious optimism on secular growth trends, cyclical factors still constrain asset returns.

• Interest rates are rising but the pace of normalization remains slow. Equilibrium yields are constrained by muted inflation, low trend growth and persistent safe asset demand.

• High global equity valuations and corporate margins reflect a mature economic cycle and are a headwind for returns. The secular outlook for stocks is reasonable as low equilibrium interest rates compensate for modest equilibrium growth in our framework, but even long-term investors must navigate the near-term cyclical challenges.

• Despite outperforming in 2017, emerging market equities remain attractive. Elsewhere, credit and real assets are bright spots, while low correlations and the start of normalization enhance alpha opportunities in private equity and hedge funds.

• Expected returns for a simple 60/40 portfolio are lower than last year, and the clockwise rotation of the stock-bond frontier—as bond returns rise and valuations constrain equity returns—is indicative of the late-cycle environment. Diversification, active allocation and manager selection in alternatives will be essential tools for managing cyclical risks in anticipation of secular growth trends that might finally be bottoming out.

J.P. MORGAN ASSET MANAGEMENT 5 EXECUTIVE SUMMARY

INTRODUCTION

As we compile the 2018 edition of our Long-Term Capital Market demographic strain remains apparent in the evolution of our global Assumptions, the world economy is enjoying its best period of growth estimates. In our 2018 LTCMAs, the outlook for global synchronized growth in more than a decade. Policymakers have growth is broadly stable at a modest level, but for the first time in coaxed an unusually long, if shallow, expansion out of the near- a decade we have cause to upgrade our growth forecast for a death experience of the global financial crisis, and despite the major economic region—the eurozone. Cyclical momentum and long-run drag of global aging, technological innovation seems to strengthening of key institutions are laying the foundation for a be at a positive inflection point—giving a tantalizing glimpse of secular uplift in trend growth, which we expect to be reinforced what might trigger a long overdue boost to productivity. with gradual but ongoing reform of labor laws. The upgrade is notable—not least in light of the eurozone sovereign debt crisis of Yet few expansions have endured so many skeptics, and few bull 2010-12—but it is not sufficient to spur a broader upgrade to markets have remained so unpopular. High levels of debt, the aggregate global growth. perceived trap of zero interest rates, the lack of inflation, and sporadic political dysfunction do more than keep some investors Against this backdrop, we have shifted our attention to a up at night; they often hijack the market narrative. Many of the more rigorous analysis of some of the secular themes—along patterns and trends in this year’s LTCMAs will be familiar to regular with a reflection on their nearer-term cyclical implications—that readers, but given the maturity of this business cycle, the interplay are set to drive asset returns in the coming decade. The themes of cyclical and secular factors is set to influence long-term may be felt directly, through their economic impact, as is the case investment outcomes far more than in recent years (Exhibit 1). for our themes of technological innovation, Chinese financial markets liberalization and long- and short- term drivers of the Since 2010, the steady trend of population aging has cut almost U.S. dollar. Others may be felt more indirectly via policy channels, half a point from our estimates of trend GDP in developed as we’d anticipate for our demographics and pension economies, and the post-financial crisis productivity slump shaved de-accumulation themes. the estimate by a further quarter point. At a global level, the steadily increasing weight of faster-growing emerging markets partially buffered the effect of aging populations, but the

Reduced long-term growth estimates have pulled down equilibrium returns for most asset classes compared with the last 25 years, but there is an additional cyclical penalty in most key assets due to extended valuations and low starting yields EXHIBIT 1: HISTORICAL 25-YEAR AVERAGE RETURNS FOR KEY ASSETS AND THIS YEAR'S ESTIMATES, SPLIT INTO THEIR SECULAR (EQUILIBRIUM) AND CYCLICAL COMPONENTS

25 year average returns 2018 cyclical 2018 secular 2018 LTCMA return estimate % 10

-2 U.S. equity U.S. high yield U.S. investment grade bonds U.S. aggregate bonds U.S. 60/40

Source: Bloomberg, Datastream, J.P. Morgan Asset Management Multi-Asset Solutions; data as of September 30, 2017.

6 LONG-TERM CAPITAL MARKET ASSUMPTIONS 2018 LONG-TERM CAPITAL MARKET ASSUMPTIONS

MACROECONOMIC THEMES: BALANCING SECULAR Demographic trends can be tweaked a little by policy—raising the AND CYCLICAL FACTORS retirement age, increasing net migration, etc.—but the economic drag of a slower-growing working age population is persistent. Our 10- to 15-year aggregate forecasts for global growth are unchanged this year, with developed markets at 1.50% and If the net economic effect of an older population is simply emerging markets at 4.50% (Exhibit 2). This translates to rather to limit potential growth, its impact on asset returns is more modest return expectations for most asset classes at equilibrium, nuanced. Lower growth drives down equilibrium interest rates, b which are further affected by the cyclical positioning of the ut a competing factor is that as populations age, “savings gluts” 1 economy and markets. Simply put, we’ve already banked a great (considered in some analyses to depress equilibrium interest deal of the current cycle’s returns, and with the global expansion rates) may start to unwind. Another important consideration is now quite mature there’s little scope for any cyclical uplift to that vast pools of pension assets are currently underfunded, with average returns. But if we look beyond the limitations that today’s lower long-term interest rates exacerbating that underfunding. mature cycle places on long-term returns, there are reasons for As pension plans seek to fulfill their commitments to a growing cautious optimism and even some upside risks emerging for our pool of retirees, the incremental demand for income from higher long-term economic forecasts. The impact of demographics has risk assets such as credit and equity may remain surprisingly been well flagged, and has led to the steady lowering of potential strong. While policymakers may find the temptation to tackle growth estimates, but the possible upside from technology and its pension underfunding by tolerating higher inflation somewhat consequent additional boost to productivity has yet to come into appealing—financial repression by another name—it overlooks that base case economic forecasts. many funds are already inflation proofed. But the bigger problem, perhaps, is that inflation isn’t what it used to be and $11 trillion2 Our long-term macroeconomic assumptions stand at the of central bank interventions since the global financial crisis have intersection of two powerful secular forces: demographics and failed, so far, to push inflation meaningfully higher (Exhibit 3). productivity. While the long-term economic impact of an aging population is well documented, estimating productivity remains 1 BIS working papers No. 656, Goodhart and Pradhan; Ben Bernanke speech at the “dark art” of economics. These secular forces tend to define Virginia Association of Economists, Richmond, Virginia on March 10, 2005. potential growth over the long run. Short-term cycles fluctuate 2 The combined balance sheets of the U.S. Federal Reserve, Bank of Japan, around this trend as demand ebbs and flows, but the supply-side European Central Bank, Bank of England and Swiss National Bank expanded from approximately USD 4.3 trillion at the end of 2007 to approximately USD 15.5 trillion factors effectively set an “average speed limit” for the economy. at the end of Q3 2017.

Our 2018 assumptions anticipate slow real GDP growth globally; neither global growth assumptions nor the emerging market-developed market growth gap has changed from last year EXHIBIT 2: MACROECONOMIC ASSUMPTIONS (%) 2018 assumptions 2017 assumptions Change (percentage points) Real GDP Core inflation Real GDP Core inflation Real GDP Core inflation DEVELOPED MARKETS 1.50 1.75 1.50 1.75 0.00 0.00 U.S. 1.75 2.25 1.75 2.25 0.00 0.00 Eurozone 1.50 1.50 1.25 1.50 0.25 0.00 UK 1.25 2.00 1.25 2.00 0.00 0.00 Japan 0.50 1.00 0.50 1.00 0.00 0.00 Australia 2.00 2.25 2.25 2.25 -0.25 0.00 Canada 1.50 1.75 1.50 1.75 0.00 0.00 Sweden 1.75 1.75 1.75 1.25 0.00 0.50 Switzerland 1.25 0.75 1.50 0.75 -0.25 0.00 EMERGING MARKETS* 4.50 3.50 4.50 3.75 0.00 -0.25 Brazil 3.00 5.00 2.75 5.25 0.25 -0.25 China 5.00 2.75 5.25 3.00 -0.25 -0.25 India 7.00 5.00 7.00 5.00 0.00 0.00 Russia 1.50 5.50 2.25 5.50 -0.75 0.00 GLOBAL 2.50 2.50 2.50 2.50 0.00 0.00 Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. * Emerging markets aggregate derived from 9-country sample.

J.P. MORGAN ASSET MANAGEMENT 7 EXECUTIVE SUMMARY

The balance sheets of the major central banks grew by over estimate of long-term fair value, the path of the dollar is a critical $11 trillion since 2007, yet core inflation remains low consideration for any international diversification throughout our EXHIBIT 3: GLOBAL CENTRAL BANKS BALANCE SHEETS SIZE ($, BILLIONS) 10- to 15-year investment horizon. VS. PREVAILING GLOBAL CORE* INFLATION BoE SNB BoJ ECB Fed Aggregate core CPI If demographics are defining much of the debate over long-term 16,000 2.5 economic risks, then it is productivity—and specifically technology— that creates the main upside risk to our outlook. Although 2.0 12,000 productivity growth seems to have stalled in the aftermath of the global financial crisis, we reject the idea that productivity is 1.5 8,000 permanently impaired. Based on the pace of development in

$, billions 1.0 automation and artificial intelligence, we estimate a potential upside risk to our baseline growth forecasts of 1.0%-1.5% if the 4,000 0.5 promise of today’s innovations are fully realized in productivity gains (Exhibit 4). 0 0.0 ’06 ’08 ’10 ’12 ’14 ’16

Source: J.P. Morgan Asset Management Multi-Asset Solutions, Bloomberg; data If just one tenth of the potential growth gain from technology through September 2017. accrues each year, it could boost G3 GDP by $6.5 trillion by the *Core CPI across U.S., Switzerland, UK, eurozone and Japan, weighted by GDP. early 2030s EXHIBIT 4: G3 GROWTH BOOST FROM TECH Population aging is a key factor in China’s long-term outlook, Total real GDP (G3) - with tech boost Total real GDP (G3) in which gradually declining trend growth and the possible reversal $55,000 in public sector savings rates could have a significant impact on global capital flows. China is the world’s second-biggest economy, representing 15% of world GDP,3 and is the largest international $50,000 holder of U.S. debt. Yet China's domestic financial markets are undeveloped and difficult to access. The portion of China's stock $45,000 market that is open to foreign investors accounts for just 3% of $, billions global free-float equity market capitalization.4 However, the total scale of the Chinese equity market—including both domestic and $40,000 non-domestic—is almost six times this size. We believe that over the next decade China's financial system will open up meaningfully to $35,000 market forces and foreign access will be considerably easier. ’18 ’21 ’24 ’27 ’30 ’33 Liberalization of the financial system could exert some upward Source: J.P. Morgan Asset Management Multi-Asset Solutions, Bloomberg, IMF pressure on Chinese equilibrium interest rates and, over time, attract World Economic Outlook; data as of September 30, 2017. Note: G3 economies defined as U.S., eurozone and Japan. global savings away from lower yielding G4 bond markets. Such secular trends have led us to publish our first set of long-term Chinese asset return estimates this year. MAJOR ASSET CLASS ASSUMPTIONS

These trends also strengthen our conviction that, over the long Our 2018 assumptions represent something of a watershed in our term, the U.S. dollar is moving lower. The reversal of fortunes in long-term economic outlook. Over the last decade, worsening 2017 highlights the cyclical and structural crosscurrents that affect demographic trends drove successive downgrades to our estimates the U.S. dollar as secular drivers such as growth differentials and of trend growth, but this sequence of downgrades may be nearing current account imbalances once again begin to dominate monetary an end. Moreover, for the first time, we can see tangible upside risks policy in dictating the currency’s path. Given that the U.S. dollar to long-term trend growth expectations starting to emerge. It is too remains above our current early to declare a trough in long-term trend growth estimates or, indeed, to fully bake in the promise of a technology-driven boost to productivity to our base case numbers. However, the possibility that 3 April 2017 IMF World Economic Outlook, estimates for world and China 2017 we are nearing the end of a prolonged sequence of downgrades to nominal GDP. long-term trend growth is significant for estimates of equilibrium 4 From MSCI data on free-float market capitalization: world, USD 43.7 trillion; China, USD 1.5 trillion. asset returns.

8 LONG-TERM CAPITAL MARKET ASSUMPTIONS 2018 LONG-TERM CAPITAL MARKET ASSUMPTIONS

Cyclical pressures are weighing on returns for long-term equity and riskier credit, while higher starting yields push bond returns modestly higher EXHIBIT 5A: SELECTED LTCMA RETURNS (%) EXHIBIT 5B: SELECTED LTCMA RISK PREMIA (%)

2018 LTCMA 2017 LTCMA 2018 LTCMA 2017 LTCMA

5.75 U.S. small cap High yield premium 1.75 7.00 2.50 5.50 0.50 U.S. large cap Investment grade premium 6.25 1.00 5.25 U.S. high yield bonds Private equity premium* 1.00 5.75 1.00 3.50 U.S. investment grade corporate Small cap premium 0.25 3.25 0.75 3.00 2.50 U.S. intermediate Treasuries Equity risk premium 2.25 4.00 2.00 U.S. cash Duration premium 1.00 2.00 0.25

Source: J.P. Morgan Asset Management, estimates as of September 30, 2016 and September 30, 2017. * Private equity premium assumptions are our alpha assumptions for private equity.

This year, though, the maturity of the current economic cycle is EQUITY—Cyclical drags, but still the best for long-term manifesting itself only too clearly in asset returns. Equity return returns expectations, especially, are displaying a cyclical drag from full Our equity return forecasts are down 0.5% to 0.75% for most valuations and above-average margins. These factors will very regions this year, reflecting primarily the strong returns delivered by likely be in place for the remainder of the current business cycle. stocks in 2017. With growth forecasts generally little changed As the cycle matures, even investors with a very long investment compared with 2017, equilibrium revenue growth levels are horizon will need to further balance cyclical and secular influences unchanged, but both margins and valuations are a constraint on on asset returns. Indeed, it will be a critical consideration future returns in most major regions. In our estimation of long-term (Exhibits 5A and 5B). returns, equities, more than other asset classes, are reflecting the FIXED INCOME—No rush to normalize interplay of cyclical considerations and secular growth projections. The equity risk premium (ERP) has fallen meaningfully vs. last year The slow and shallow path of normalization that we discussed last but remains close to its long-run average and is distinctly attractive year is emerging as the preferred central bank playbook. Quite when compared with the duration premium. Nevertheless, the simply, the path of rates we described in our 2017 LTCMAs is little principal source of returns from developed market (DM) equities changed, bar rolling forward by a year. Return forecasts for G4 remains dividends and buybacks, and with this now a persistent sovereign bonds are slightly improved this year, given the modest trend, it is likely to push investors seeking capital growth further rebound in the duration premium as G4 bond yields have pushed toward EM equities and private equity markets. up from their lows. Nevertheless, the outlook for most government bond returns remains challenged and only ALTERNATIVE ASSETS—Alpha better, beta abating moderately above that of cash. Credit continues to offer a decent Lower equity return forecasts naturally create a headwind for pickup in returns even though prevailing spread levels are tighter private markets, but the downtrend in alpha expectations of than our estimates of fair value. The duration component of recent years looks to have taken a pause as monetary policy corporate bonds is also a factor in the modest changes to return around the world starts to normalize. High valuations are still an forecasts—investment grade returns up 0.25% and high yield impediment to private equity returns, but at the same time, lower returns down 0.50%. The outlook for emerging market (EM) prevailing correlation across equity markets and a widening economies continues to improve, and greater structural stability opportunity set serve to increase potential for alpha and should offset some of the lingering fears over leverage. As a enhanced returns. The improvement in alpha opportunities, result, EM debt continues to offer fixed income investors both though, is more pronounced in the hedge fund sector, where the diversification and reasonable return potential. shift from a macro-driven to a more fundamentally driven investing environment is supportive. Improved alpha opportunities

J.P. MORGAN ASSET MANAGEMENT 9 EXECUTIVE SUMMARY and worsening beta underline our view that manager selection Forecast Sharpe ratios for bonds are well below their long-term remains the critical driver of returns for both private equity and average hedge funds. Indeed, we anticipate a significant premium to our EXHIBIT 6: RISK-ADJUSTED RETURN ASSUMPTIONS VS. HISTORICAL AVERAGES ACROSS ASSET CLASSES (SHARPE RATIOS) long-term return estimates for upper quartile managers. 2018 LTCMA Historical (June 30, 2006-June 30, 2017) We see a relatively attractive outlook for real asset returns, and U.S. 10-yr Treasuries 0.12 despite the advanced stage of the U.S. business cycle we expect 0.73 U.S. long Treasuries 0.04 that real estate will prove to be quite resilient. Our estimates fall 0.47 U.S. investment grade corporate bonds 0.25 modestly, but supply and leverage constraint mean that the typical 0.85 “late-cycle exuberance” that so often appears in real estate U.S. high yield bonds 0.38 markets is largely absent. More broadly, real assets remain well 0.71 Emerging markets sovereign debt 0.33 supported by a combination of the strong and persistent bid for 0.80 AC world equity 0.26 such assets and the relatively high illiquidity premiums that are 0.28 keeping long-term return estimates elevated. U.S. large cap 0.25 0.51

EAFE equity 0.25 FOREIGN EXCHANGE—Secular forces starting to prevail 0.12 Emerging markets equity 0.28 The secular forces that we expect to nudge the U.S. dollar down 0.19 over the longer term began to assert themselves in 2017, when Source: J.P. Morgan Asset Management Multi-Asset Solutions; estimates as of they started to dominate the cyclical monetary policy that had September 30, 2017. driven the dollar’s appreciation in 2014-15. Despite the sharp decline in USD in 2017, the currency remains meaningfully above Last year we argued that, after years of borrowing returns from our long-term estimates of fair value, which we see as 1.34 for the future, the future had finally arrived. This remains our view EUR/USD and 93 for USD/JPY. Now that the dollar has retreated today. As we look forward, cyclical factors loom large, even for from its recent highs,we would expect the pace of further declines long-term investors, but the secular trends of lower growth and to moderate somewhat. But the impact of currency translation on lower equilibrium interest rates must also be considered. Lower asset returns in globally diversified portfolios will continue to be equilibrium interest rates and the glacial pace of normalization an important consideration for many investors. put secular constraints on government bond returns that will likely transcend this cycle. Even if trend growth surprises positively and IMPLICATIONS FOR INVESTORS equilibrium rates rise, they will force losses on government bondholders as yields adjust from cyclically depressed levels. Risk-adjusted returns for fixed income assets now stand at he other side of this is that lower interest rates, all else equal, the low end of their typical ranges (Exhibit 6). As last year, exert upward pressure on equity risk premia and valuations, which equity Sharpe ratios are substantially better than those for in turn may increase the lower bound for equity multiples over the government bonds, but in absolute terms U.S. large cap Sharpe full business cycle. ratios are below the average of the last 10 years, while EAFE and EM equity Sharpe ratios are above their 10-year average. Subdued The asset market impact of the structural themes we explore in this forecasts for equilibrium growth play some part in this. But it is paper is likely to be both complex and nuanced. Aging populations also a function of how late we are in the economic cycle, and of are an impediment to growth but may equally start to reverse the valuation headwinds that most regions face. Further cyclical savings gluts. Technological innovation could meaningfully boost momentum could certainly front-load returns from equities—much trend growth but is likely to be disinflationary at the margin. On as it has in 2017—but long-term investors should be clear that this balance, we expect modest baseline economic growth with some is a cyclical, not a secular, trade. Overcoming the pressure on upside risk from productivity gains, but generally subdued inflation. long-term potential returns from full valuations and high margins This condition should not necessitate overtly tight policy, reinforcing will require more active positioning around long-term themes the secular anchor of low equilibrium rates. For investors of all (e.g., technology, Chinese financial liberalization, etc.), greater use types, recognizing this long-term trend will be essential in of alternatives, an element of timing—or, most likely, all three. calibrating how today’s cyclical environment will normalize and how Despite these challenges for investors, the incremental to construct robust and versatile portfolios. deterioration in risk-adjusted returns from 2017 is quite marginal, and opportunities to enhance returns and generate alpha persist.

10 LONG-TERM CAPITAL MARKET ASSUMPTIONS 2018 LONG-TERM CAPITAL MARKET ASSUMPTIONS

This year, our expected return for a U.S. dollar-based 60/40 But opportunity lies in the clustering of many asset classes close portfolio is slightly lower at 5.25%, down from 5.50% last year, to, and even above, the stock-bond frontier, which implies ample and the stock-bond frontier has rotated clockwise (Exhibit 7). This scope for diversification and enhancing returns.5 Beyond this is a clear manifestation of the cyclical and secular considerations current cycle, however, there is reason for cautious optimism, that investors face: Modest baseline growth continues to limit as the persistent pattern of declining long-term trend growth long-term return expectations for static balanced portfolios, while estimates might be nearing an end. Indeed, the outlook this year the late-cycle conditions of elevated equity valuations and for long-term investors might well be somewhat brighter than the gradually rising interest rates flatten the stock-bond frontier. The dip in return estimates may imply. challenge for investors is that returns for a simple, static 60/40 portfolio are unlikely to turn meaningfully higher until this cycle 5 We use a simple, two-asset stock-bond frontier; addition of further assets— ends—and that means riding out an inevitable period of disruption. particularly uncorrelated ones—will typically improve the efficiency of the resulting portfolio.

Compared with last year, 60/40 returns are a little lower, and the late stage of the current cycle is rotating the stock-bond frontier in a clockwise direction; although returns are subdued, the clustering of many assets close to the frontier implies ample opportunity for diversification EXHIBIT 7: USD STOCK-BOND FRONTIERS AND 60/40 PORTFOLIOS BASED ON 2018 VS. 2017 LTCMAS FOR RISK AND RETURN (%)

2018 stock-bond frontier 60/40 Portfolio (2018) 2017 stock-bond frontier 60/40 Portfolio (2017) 9

8 Emerging markets equity Private equity 7

60/40 PORTFOLIO (2017) 6 EAFE equity U.S. high yield 5 U.S. large cap AC world equity U.S. direct real estate 4 Emerging markets sovereign debt Diversiﬁed hedge funds Commodities U.S. aggregate bonds Compound return (%) 3 60/40 PORTFOLIO (2018) U.S. intermediate Treasuries 2 U.S. cash

0 0 5 10 Volatility 15 20 25

Source: J.P. Morgan Asset Management; estimates as of September 30, 2016 and September 30, 2017.

J.P. MORGAN ASSET MANAGEMENT 11 MACROECONOMIC ASSUMPTIONS

Mostly stable, mostly moderate Michael Hood, Global Strategist, Multi-Asset Solutions Dr. David Kelly, CFA, Chief Global Strategist, Head of Global Market Insights Strategy

IN BRIEF • This year’s edition of our Long-Term Capital Market Assumptions (LTCMAs) contains few significant changes to projected real GDP growth and inflation in major economies. The main themes from recent years’ forecasts remain in place.

• Real growth will run at a modest pace by long-term historical standards, with aging populations representing the most important source of slowing relative to the past.

• The U.S. will continue to record somewhat stronger growth than the euro area, UK and Japan.

• As a group, emerging market (EM) economies will grow faster than their developed market (DM) counterparts. India will lead the way in emerging market growth as China gradually slows.

• Middle-of-the-road inflation, in general fairly close to central bank targets, will prevail in the long run.

12 LONG-TERM CAPITAL MARKET ASSUMPTIONS MOSTLY STABLE, MOSTLY MODERATE

MILDLY STRONGER GROWTH IN EUROPE, SLIGHTLY before). The Chinese economy has been decelerating gradually, WEAKENED OUTLOOK FOR CHINA a trend we expect to continue, so that rolling our projections forward by a year brings down the full-period average. We have The 2018 edition of our Long-Term Capital Market Assumptions raised our GDP forecasts for Brazil (up 0.25ppt to 3.00%) and makes few significant changes to our projections for real GDP South Africa (up 0.50ppt to 2.75%) while cutting our figures for growth and inflation in major economies. We have nudged up our Russia (down 0.75ppt) and Taiwan (down 0.25ppt). Some of the euro area real GDP projection to 1.50% (from 1.25% in our 2017 movement in these EM projections reflects an improved edition), while trimming our Australia and Switzerland forecasts by understanding of human capital dynamics, which look set to help a quarter point each, to 2.00% and 1.25%, respectively. The upward Brazil and South Africa over the long run while undermining revision to the euro area GDP number reflects greater optimism Russian growth. A number of the EM inflation projections have about trend growth and today’s starting point. We think that moved by a quarter point, notably an upward shift in South Africa structural reforms pursued in the euro area in recent years will and Turkey, where we observe less diligence than previously in likely boost the underlying growth trend through two channels: pursuit of central bank targets. Our projections for emerging labor force participation and total factor productivity (TFP). markets in aggregate are 4.50% real GDP growth, the same as Meanwhile, the region’s currently elevated unemployment rate last year, and 3.50% inflation, down 0.25ppt, pointing to 8.00% should allow it to expand at an above-potential clip in coming years. nominal GDP. On balance, our DM real GDP aggregate holds steady at 1.50%. Along with an unchanged set of inflation projections (at 1.75% for the group as a whole), our DM nominal growth assumption remains GDP GROWTH: DEMOGRAPHIC AND PRODUCTIVITY at 3.25% (Exhibit 1). DRIVERS

We have made several modifications to our emerging market GDP To construct our real GDP forecasts, we estimate a long-term trend and inflation assumptions. Most important, we are lowering our or potential rate of growth for each economy, focusing on supply- expectations for Chinese nominal growth by half of a percentage side drivers. Potential growth thus stems from (1) the labor input, point (ppt) from our projection in 2017: Half of the decrease itself divided into employment, average hours worked and human reflects a slight drop in real growth (to 5.00%, vs. 5.25% last capital; (2) capital stock expansion, reflecting investment spending; year), and half comes from inflation (now 2.75%, vs. 3.00% and (3) total factor productivity, a residual that in developed

Our 2018 assumptions anticipate slow real GDP growth globally; neither global growth assumptions nor the EM-DM growth gap has changed from last year EXHIBIT 1: MACROECONOMIC ASSUMPTIONS (%) 2018 assumptions 2017 assumptions Change (percentage points) Real GDP Core inflation Real GDP Core inflation Real GDP Core inflation DEVELOPED MARKETS 1.50 1.75 1.50 1.75 0.00 0.00 U.S. 1.75 2.25 1.75 2.25 0.00 0.00 Eurozone 1.50 1.50 1.25 1.50 0.25 0.00 UK 1.25 2.00 1.25 2.00 0.00 0.00 Japan 0.50 1.00 0.50 1.00 0.00 0.00 Australia 2.00 2.25 2.25 2.25 -0.25 0.00 Canada 1.50 1.75 1.50 1.75 0.00 0.00 Sweden 1.75 1.75 1.75 1.25 0.00 0.50 Switzerland 1.25 0.75 1.50 0.75 -0.25 0.00 EMERGING MARKETS* 4.50 3.50 4.50 3.75 0.00 -0.25 Brazil 3.00 5.00 2.75 5.25 0.25 -0.25 China 5.00 2.75 5.25 3.00 -0.25 -0.25 India 7.00 5.00 7.00 5.00 0.00 0.00 Russia 1.50 5.50 2.25 5.50 -0.75 0.00 GLOBAL 2.50 2.50 2.50 2.50 0.00 0.00 Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. * Emerging markets aggregate derived from 9-country sample.

J.P. MORGAN ASSET MANAGEMENT 13 MACROECONOMIC ASSUMPTIONS economies primarily owes to technological change. We forecast We assume that total factor productivity contributes 0.6ppt per each of the subcomponents separately, combining them by using year to U.S. real GDP growth during our forecast horizon, primarily the shares of each country’s output that come from labor and representing the advancement in the global technology frontier. capital. This approach allows for more granularity than in our This pace would represent a small pickup from the recent trend— previous work, which relied on broader concepts of labor force with the world seemingly having gone through a fallow period for and labor productivity growth. technological change in the past decade—but would lie noticeably below the long-term historical average. TFP forecasts for other DM economies, which also generally operate at the technology DEVELOPED MARKET GROWTH: U.S. LEADS; EURO frontier, cluster tightly around the U.S. figure, with Switzerland AREA IN RELATIVELY FAVORABLE POSITION and Canada—countries that have relatively sluggish TFP histories and face significant resource reallocation needs in the coming Demographic forces, as reflected in the employment part of the years—a bit below. labor input, explain much of the expected shortfall in economic growth compared with history. Differences in this factor also Standard growth theory creates an expectation that economies generate the lion’s share of dispersion among DM economies will expand in a balanced way, with the capital stock rising more (Exhibit 2). We work off year-by-year population projections made or less in line with the other inputs (labor and TFP). In practice, by the U.S. Census Bureau, focusing on the 15–64 and 65-plus age this prediction has broadly held true, as the capital-to-GDP ratio cohorts. We then run several simulations for each country, with has essentially held steady over time in DM economies (Exhibit 3). varying paths for labor force participation in these two age groups. In our assumptions, this balance persists. We project DM capital This exercise gives us a plausible range for labor force growth, and stocks to grow at a similar pace to the combination of the labor we generally settle in the middle of this band in making our input and TFP. forecast. Projections for average hours worked and human capital improvement vary only marginally across DM economies, although Growth in most economies has been balanced over time, with the the euro area gets a small fillip relative to others in human capital, capital stock rising in line with other inputs thanks to ongoing gains in educational attainment in Southern EXHIBIT 3: CAPITAL STOCK AS A RATIO TO GDP (X) European countries. For developed economies as a whole, though, U.S. JP AU SE human capital represents another source of slowdown compared EMU UK CA CH with the past, as the multi-decade trend toward longer schooling X 6.0 has slowed significantly in recent years.

5.0 Demographic factors account for much of the dispersion among expected GDP growth rates in DM economies 4.0 EXHIBIT 2: PROJECTED EMPLOYMENT GROWTH (%, ANNUAL AVERAGE)

0.7 3.0 0.6 0.5 2.0

1.0 ’90 ’96 ’02 ’08 ’14 0.2 0.2 0.1 Source: Penn World Table, J.P. Morgan Asset Management; data as of September 30, 2017. U.S. UK AUSTRALIA SWEDEN SWITZERLAND CANADA

-0.1 JAPAN EUROZONE

-0.4

Source: U.S. Census Bureau, J.P. Morgan Asset Management; data and forecasts as of September 30, 2017.

14 LONG-TERM CAPITAL MARKET ASSUMPTIONS MOSTLY STABLE, MOSTLY MODERATE

These inputs lead us to an expected rank order for the G4 The starting level of human capital varies widely across EM economies, with the U.S. leading the way, followed by the euro economies and should provide a relative boost to previous laggards area, UK and Japan. As mentioned, some of our relative optimism as educational attainment rises about the euro area stems from the economy’s cyclical position, EXHIBIT 4: ESTIMATED HUMAN CAPITAL LEVELS* which looks more favorable than the others, along with a 3.6 projection that total factor productivity growth will match that of 3.4 the U.S., having generally lagged by a small amount in the past. We continue to assume that the currency union will survive in 3.2 something similar to its present form throughout our forecast horizon, and we note that political threats to the euro area appear to have receded since last year’s edition. Long-term prospects for 2.7 2.7 2.7 the UK remain cloudy as the country negotiates its exit from the European Union. Our GDP projection for the UK, which already 2.5 embeds slower TFP gains than in other major DM economies, has 2.3 not changed this year, but we believe the risks to the outlook are tilted to the downside, particularly if immigration policy 2.1 ultimately compromises one of the UK’s advantages, which is a BRAZIL CHINA INDIA KOREA MEXICO RUSSIA TURKEY TAIWAN SOUTH AFRICA relatively favorable demographic situation. Source: Penn World Table, J.P. Morgan Asset Management; data as of September 30, 2017. * Estimated human capital levels are as measured by the PWT human capital index, a measure of human capital per person based on average years of schooling and assumed EMERGING MARKET GROWTH: EXCEEDING THE DM rate of return to education. PACE, THOUGH NOT EVERYWHERE Despite China’s impressive advance over the past 20 years, its per We build our forecasts for EM economies through the same capita income level remains below the EM average. As mentioned, process. The faster rate of expected growth for the EM group as a we believe ongoing urbanization will continue to boost the country’s whole owes primarily to three factors. First, while facing their own labor force, and China has generated fairly steady TFP gains over demographic challenges, EM countries are still experiencing faster time, though the latter has slowed recently. More uncertainty increases in population, and by extension labor forces, than their surrounds the China forecast than some others, however, for two DM counterparts. That said, the outlook on this front varies widely reasons. First, China’s rapid development after 1990 depended across EM economies, with some—notably Russia and Taiwan— heavily on very strong investment spending, and its capital stock likely to experience labor force shrinkage in coming years. The rose rapidly relative to GDP, in contrast to the global norm. With the China story is more nuanced. While its prime-age population will capital stock now fairly elevated, we expect more balanced growth decrease over the next 15 years, China will likely benefit from from China in the future, but the country’s ability to find a new ongoing urbanization, given that it remains relatively rural for its economic model remains far from assured. Second, a steep increase stage of development, even after heavy city-bound migration in in leverage has accompanied China’s development, leaving the recent decades. Second, in comparison with DM economies, EM country vulnerable to a financial setback of some sort in coming countries should experience greater improvement in the quality of years, though the timing of such an event is essentially impossible their labor forces as school attainment continues to rise for most to predict. We are comfortable projecting that over the next 15 of the group (Exhibit 4). Third, TFP growth will likely continue to years China will grow significantly more slowly than during the past exceed the DM pace in some, though not all, EM economies, 15 but faster than the EM average, while recognizing a wide especially in the manufacturing- and tech-oriented EM Asian uncertainty band around our point forecast. At the other end of the economies with proven track records of convergence toward the spectrum, Korea and Taiwan, which have nearly reached DM global frontier. economy status in per capita income terms, will likely grow only Our EM forecasts embed a roughly inverse relationship between marginally faster than developed market economies. We see Russia, current per capita income and expected future growth. India, the with its uniquely poor demographic outlook and commodity- poorest country in our sample, leads the pack; we project strong dependent economy—unlikely to attract any great burst of capital gains in the labor force, human capital, TFP and (by extension) investment absent another surge in price of oil—bringing up the capital stock. China comes in second on our list. rear among EM countries.

J.P. MORGAN ASSET MANAGEMENT 15 MACROECONOMIC ASSUMPTIONS

INFLATION: RANGE-BOUND AND CLOSE EMERGING MARKET INFLATION: WIDE DISPERSION, TO TARGETS THOUGH TURMOIL UNLIKELY

We use central bank targets as the starting point for our inflation For EM economies, we follow the same approach, forecasts. At the same time, we take into account our views about while bearing in mind several key differences from the DM policymakers’ commitment to those goals, as well as historical atmosphere: (1) in some cases, vaguer or non-existent inflation track records and any idiosyncratic forces that might affect a targets; (2) central banks that may be less formally independent country’s inflation rate over the long run (such as the likelihood of or more subject to political influence; (3) shorter track records of large and persistent currency moves). Several beliefs about success; and (4) inflation indices that place greater weight on inflation underlie our projections. First, our reading of history goods prices—which, though their inflation rates tilt lower than suggests that monetary policy works: When central banks pursue services prices, are also more prone to buffeting by the exchange clear inflation targets using all available tools, they generally rate swings prevalent in emerging markets. For these reasons, our come close to their aims. Second, barring extraordinary EM inflation forecasts generally lie above these countries’ central circumstances—such as a central bank losing control over its own bank targets. That said, we do not expect a return to the instability balance sheet—inflation tends to run in fairly narrow channels. that characterized many EM economies before the 1990s. For the The global experience since the financial crisis provides instruction countries in our sample, institutional improvements and voter- in this respect. Despite very large and sustained output gaps, supported commitments to fairly low inflation make a recurrence no major economy sank into persistent deflation in recent years. of such turmoil unlikely. On the other hand, no DM country saw inflation skyrocket, even as central banks have pursued extraordinary accommodative policy As is the case with our emerging market GDP projections, our EM stances for an extended period of time. Third, again in line with inflation projections show greater dispersion than those for DM historical experience, we see no long-run trade-off between countries. At the low end of the spectrum, we forecast 1.25% for growth and inflation. In other words, we do not expect changes to Taiwan, which lacks a formal inflation target but has recorded individual countries’ GDP and consumer price index (CPI) forecasts quite low inflation during the past decade. By contrast, we pencil to be correlated. in 7.00% for Turkey, which has averaged slightly over 8% in the past 10 years. Although the Turkish central bank maintains an inflation target (currently 5%), it has not shown any great willingness to DEVELOPED MARKET INFLATION FORECASTS: sacrifice growth in the short run to hit its goal, and we view UNCHANGED FROM LAST YEAR monetary policymakers in Turkey as subject to significant political pressure. The other EM economies range between these two. We Given those views, and absent any major surprises in the inflation forecast 2.75% for China, similar to its 10-year average, but environment since last year’s LTCMAs, our DM inflation forecasts acknowledge considerable uncertainty around that estimate, with have not changed this year. As in the past, we project 2.25% CPI the risk distribution tilted toward the high side. With elevated inflation in the U.S., consistent with the Federal Reserve meeting leverage in its economy, China could conceivably require a period of its 2% target for the index it follows, the consumption spending somewhat higher inflation to make its debt burdens sustainable, deflator (for which the inflation rate tends to run a bit below the and its move to a more services-oriented economy also will likely CPI). The asymmetric nature of the European Central Bank’s 2% put some upward pressure on overall inflation, given slower inflation target, and its historically hawkish bias, leave our euro productivity improvement in that sector. Chinese inflation has area inflation projection at 1.75%. Japan’s ongoing efforts to boost averaged a modest 2.0% over the past five years, however, so our inflation from its currently low level put our forecast at 1.00% for forecast already points to a pickup from recent trends. the 15-year period, with an expectation of a 1.50% to 1.75% inflation rate at the end of our horizon.

16 LONG-TERM CAPITAL MARKET ASSUMPTIONS I Thematic articles TECHNOLOGY, PRODUCTIVITY AND THE LABOR FORCE

TECHNOLOGY, PRODUCTIVITY AND THE LABOR FORCE

The impact of technology on long-term potential economic growth John Bilton, CFA, Head of Global Multi-Asset Strategy, Multi-Asset Solutions Shrenick Shah, Portfolio Manager, Multi-Asset Solutions Patrik Schöwitz, CFA, Global Strategist, Multi-Asset Solutions Michael Albrecht, CFA, Global Strategist, Multi-Asset Solutions Brian Bovino, Associate, Multi-Asset Solutions

IN BRIEF • Technological change is advancing with unprecedented speed, scope and scale—and with potentially far-reaching effects across economies and societies. Amid explosive growth in processing power and machine learning, a world where artificial intelligence (AI) eventually rivals human intelligence can no longer be dismissed as science fiction, with profound implications for economic growth and the labor force. • Technology will affect economic growth rates and capital market returns in ways that are difficult to foresee. Workforce automation and AI have the potential to deliver significant overall productivity gains, and some nations facing growth challenges from aging populations could see an additional boost to trend growth rates. This suggests a possible increase to our current Long-Term Capital Market Assumptions (LTCMA) estimate of trend GDP growth in developed economies of 1% to 1.5%, while narrowing the growth spread among them. • However, the latent power of automation has also raised fears of substantial job losses. Historically, fears of technological unemployment have not materialized, but, given the nature of the current wave of innovation, we cannot take historical patterns entirely for granted. If displaced workers are not efficiently re-employed, it could weigh on consumption and economic growth. Therefore, the above numbers represent a reasonable upper bound to potential growth forecasts provided displaced labor is rapidly reabsorbed. • Recent backlashes against globalization point to the social and political strains that must be addressed to fully harness the benefits of technology. We envision a changing role for governments in helping workers and the broader economy adapt to technological change. Protecting the purchasing power of consumers will be critical. • We identify five areas where we believe the early effects of technological change on the world economy are investible today: cloud computing, the Internet of Things, artificial intelligence, robotics and blockchain technology.

18 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE IMPACT OF TECHNOLOGY ON LONG-TERM POTENTIAL ECONOMIC GROWTH

INTRODUCTION significantly, it is unclear whether a modern economy that is rapidly adopting automation and AI can deliver rising wages and rising Amara’s law states that “we tend to overestimate the effect of productivity simultaneously. This creates complications in estimating technology in the short run and underestimate the effect in the the potential boost to trend GDP and in harnessing the positive—and 1 long run.” Over the last few years, the media have been saturated mitigating the negative—effects across economies and societies. with articles, papers, books and videos exploring how technology While we are probably a long way from a world in which artificial will change our lives. Some scenarios describe a fundamental intelligence rivals human intelligence, we expect today’s transformation of society, the workforce and, indeed, the world technological revolution to spark far-reaching economic, social and order that would rival even the most fanciful science fiction. geopolitical changes—perhaps eventually redefining the role of human labor in the workforce. Among many groundbreaking technologies on the horizon,2 we focus here on automation and artificial intelligence. Exponential growth in these technologies will profoundly change economies WHY NOW? and societies, and in ways we have not yet imagined. We expect Any analysis of the potentially disruptive impact of technology the impact of these changes to accrue gradually at first and that quickly runs into the question “Why now?” There are many who— Amara’s law will be proven once more. Nevertheless, we believe quite rightly—point out that the economy and labor force proved that early effects of technology on economic growth, labor, policy remarkably good at adapting to disruptive technology over the and trade will begin to be felt over our 10- to 15-year investment last 200 years. The first two industrial revolutions took the world horizon and that the early implications of those effects should from an era reliant on human and animal power to execute influence investment choices now. physical tasks into an era in which machines powered by natural In the following pages, we examine how technological change— resources provided the physical power and humans increasingly particularly automation and AI—might affect the way economies added value with their minds. New industries and functions arose grow. We structure our analysis as follows: to demand labor, such that productivity and real wages grew in tandem. However, the disruptive potential of today’s automation • We first ask “Why now?” What is it about the nature of prevailing and AI is, in our view, something altogether new. Put simply, technological advances and the shape of today’s global economy technology is now automating brains as well as brawn. that pose unique challenges? Specifically, we look at how the current situation may differ from previous The speed of this technological progress is well characterized by technological revolutions in its speed, scope and scale. Moore’s law,3 whose spirit is alive and well today in the continued exponential growth in processing speed, data storage capacity and • We then explore how technology may affect growth in connectivity relative to cost. For example, a 2017 smartphone has productivity and, ultimately, real GDP. more processing power than Cray-2, the world’s most advanced • We address the challenges to the labor force, consumption and supercomputer in 1985—a time when just two of the popular government policy that will arise from automation and AI. contemporary video game consoles were together more powerful than the computers that put man on the moon in 1969. Moreover, • We assess the implications of technological change for our the capacity for data storage has reached a critical level and LTCMAs. continues to grow exponentially, even as human education and skills • In a separate section, we explore current investment development remain linear (Exhibits 1A and 1B). In our view, data is opportunities related to automation and AI. to the information economy rather like oil is to the industrial economy—thus, the cheap, widespread and instantaneous The current wave of technological change, we conclude, availability of data, and the power to process it, are critical enablers is unlike any that has come before. Its unprecedented speed, scope of the growth in automation and AI. and scale will profoundly, and simultaneously, impact many sectors of the economy. In the industrial revolutions of the past 200 years, the economy and labor force adapted positively to disruptive technology. But we cannot assume that today’s (and tomorrow’s) workers displaced by technology will be rapidly or easily redeployed in new functions. Although technology could boost productivity

3 Moore’s law is the empirical observation that the number of transistors in an 1 Roy Amara, scientist and futurist (1925–2007). integrated circuit—closely related to computational performance—has for several 2 To name but a few others: biotechnology, nanotechnology, alternative energy. decades doubled approximately every two years.

J.P. MORGAN ASSET MANAGEMENT 19 TECHNOLOGY, PRODUCTIVITY AND THE LABOR FORCE

Human development and education follow a linear progression, while both data availability (proxied by storage costs) and the processing capabilities of machines (proxied by leading computer chess scores) are increasing exponentially EXHIBIT 1A: G7 AVERAGE YEARS OF SCHOOLING VS. DATA STORAGE COSTS EXHIBIT 1B: AVERAGE SCORES, HUMAN CHESS GRANDMASTERS VS. (GIGABYTE PER $1.00) COMPUTER CHESS PROGRAMS

G7 average years of schooling GByte per $ (log scale, RHS) Human high score Computer high score (best ﬁt) Years % 4,000 16 100,000 10,000 1,000 3,000 12 100 10 1 8 2,000 0.1 0.01 4 0.001 1,000 0.0001 0.00001 0 0.000001 0 ’50 ’65 ’80 ’95 ’10 ’25 1895 1915 1935 1955 1975 1995 2015

Source: Barro and Lee (2013), “A new data set of educational attainment in the world, Source: Chessmetrics, L. Muehlhauser, Wikipedia. 1950-2010,” Journal of Development Economics; statisticbrain.com. Data to December 2016; trends extrapolated for forward estimates.

Despite promising advances across a range of new technologies, In many cases, the impact of new technology on productivity might productivity growth seems to have stalled in the aftermath of the have been foreseen in advance, with technologies in the pipeline for financial crisis, leading many experts to reduce their optimism for some time before they had a measurable effect. For instance, Ford’s future productivity growth and thus GDP growth—a topic we Model T was introduced in 1908, but it would be about 20 more explored in the 2017 edition of our LTCMAs. We agree that years before 50% of American households owned an automobile. productivity growth is not guaranteed by some “magic force,” but The automobile, of course, had profound and measurable impacts disagree with the notion that the heydays of productivity growth on productivity. are behind us. That view is likely to be overly pessimistic: By contrast, in the past decade many technological advances, from Historically, productivity growth has come in fits and starts, often social media to the streaming of films and music, have resulted in but not always coinciding with the widespread adoption of specific large consumer benefits far exceeding the amounts paid, and their previous breakthroughs—some of which have greater impacts on impact therefore has probably not been fully captured in GDP. In productivity than others (Exhibit 2). our view, the innovations in today’s pipeline suggest meaningful

Throughout history, U.S. productivity growth has alternately surged and stalled, as the persistent and exponential growth in processing power has overlapped with both peaks and troughs in productivity—suggesting we should not extrapolate recent weakness in productivity data

EXHIBIT 2: U.S. LABOR PRODUCTIVITY GROWTH, 1950-PRESENT, AND SELECTED NOTABLE TECHNOLOGY DEVELOPMENTS

U.S. labor productivity growth Processing power %, 5-year annualized Calculations per second per constant U.S. dollar 4.0 1E+12 Facebook expands beyond 1E+11 Apollo 11 lands on colleges (’06) 1E+10 the moon (’69) Microsoft releases 1E+09 Windows (’85) 3.0 Potential upside 1E+08 1E+07 1E+06 100,000 Apple releases 2.0 iPhone (’07) 10,000 1,000 Intel introduces LTCMA forecast 100 commercial 10 1.0 microprocessor (’71) 1 Google.com registered (’97) 0.1 0.01 First web browser Amazon.com sells released to public (’91) ﬁrst book (’95) 0.001 0.0 0.0001 ’50 ’60 ’70 ’80 ’90 ’00 ’10 ’20 ’30

Source: Bureau of Labor Statistics, Goldman Sachs, Singularity.com; data as of August 2017.

20 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE IMPACT OF TECHNOLOGY ON LONG-TERM POTENTIAL ECONOMIC GROWTH opportunities to boost growth and GDP, as explored in the next These calculations apply primarily to replacing labor rather than section. Nevertheless, the potential upside from new technology complementing existing jobs or creating entirely new ones. Other brings with it concerns about labor displacement—an echo of fears technologies known to be in development—including advances in raised at technological turning points throughout history. Previous nanotechnology and bioengineering—could precipitate an entirely technological advancements were absorbed to the benefit of capital new wave of even greater productivity gains and potentially owners and the labor force alike, but the speed, scale and scope of deeper disruption, including through these two channels. They automation and AI create the unique policy challenge of optimizing may also create new opportunities for displaced labor, particularly both the level and spread of the economic gains. as the pace of adoption of new technologies today is much faster than it has been historically (Exhibit 3). ESTIMATING THE POTENTIAL UPSIDE FOR GROWTH The pace of adoption of recent technological innovations is speeding up compared with history, offering potential for new job functions In the past, technological innovation transformed society and to emerge increased labor productivity in three key ways: replacing existing EXHIBIT 3: U.S. TECHNOLOGY, RATE OF ADOPTION (%)* workers with machines and thus producing at least the same 80 output with fewer workers (e.g., refrigeration vs. the iceman); Flush toilet complementing existing workers’ jobs, boosting output per worker by automating some of their tasks (e.g., power tools); 60 or creating entirely new, higher productivity industries (e.g., Telephone Air travel computer software engineering), offsetting the displacement of ICE automobile Vacuum cleaner Home air conditioning workers by machines or replacing altogether industries that had 40 been made obsolete. Electric power Microcomputer (PC) Internet Social media* 20 Household refrigerator Tablet* Many commentators focus on labor replacement, as it may be the Years to 75% adoption rate Radio Microwave most directly measurable impact of new technology—not to mention Smartphone Cellular phone the subject of intense media attention and public debate. Driverless 0 1860 1900 1940 1980 2020 vehicles are a prominent example, offering a large potential upside Commercially available to aggregate economic output from redeploying the nearly 5 million U.S. employees operating trucks, taxis and other ground Source: Asymco, compiled from various sources with support of the Clayton Christensen Institute. transportation. If just half of these jobs were automated over the * Estimated from current adoption trends. next 20 years and, critically, displaced workers were efficiently re-employed elsewhere at average productivity, the incremental boost to GDP from driverless car automation alone would be almost RELATIVE WINNERS AND LOSERS, AND GROWTH 0.1%. It is also plausible that transportation volumes would increase, given lower costs. Of course, the assumption that labor is ACCOUNTING redeployed is central to the positive case—and some skeptics We turn now to a discussion of our Long-Term Capital Market question whether this is possible when other comparably skilled Assumptions. At the heart of our LTCMA process are 10- to 15- jobs are also being automated. But we will ignore this for one year real GDP forecasts for major economies, based on a growth moment and extend the replacement concept across the whole U.S. accounting framework (discussed below) that estimates potential economy. If we assume the most extreme estimates of automation— growth over long periods by focusing on the supply side of the such as Frey and Osborne’s projection that 47% of jobs are economy—that is, productive potential. That potential is divided computerized4—and make the exceedingly optimistic assumption into two components: total hours worked and labor productivity. that all of these individuals are redeployed into the workforce at Labor productivity is further subdivided into capital, skills and a average productivity, it would imply as much as a 3.5% per annum 6 boost to GDP growth over 20 years. More moderate estimates—for residual total factor productivity (TFP). Technological progress example, studies from the McKinsey Global Institute and PwC— drives GDP growth in developed economies most directly through suggest GDP gains from automation of approximately 1% to 1.5%.5 TFP. In our LTCMAs, we assume that all developed market (DM) economies are operating at the “technological frontier,”

4 Carl Frey and Michael Osborne, University of Oxford, “The future of employment: with the latest business practices and technology, such that TFP How susceptible are jobs to computerisation?” (September 17, 2013). 5 The McKinsey Global Institute estimates 0.8% to 1.4%; PwC sees a 14% boost to 6 This framework is analogous to the widely used Cobb-Douglas production function, world GDP by 2030, approximately 1.5% per annum. which represents output given two or more inputs (e.g., capital, labor).

J.P. MORGAN ASSET MANAGEMENT 21 TECHNOLOGY, PRODUCTIVITY AND THE LABOR FORCE advances at roughly the same pace across all the DM economies. Automation is a clear boost to total factor productivity growth, but Emerging market (EM) economies, in contrast, are still catching up the related labor impact must also be taken into account in to the technological frontier and thus should experience faster estimating trend growth rates TFP growth. EXHIBIT 4: POTENTIAL IMPACT OF TECHNOLOGY ON GDP GROWTH, TREND GROWTH ESTIMATES, NEXT 15 YEARS

Automation: The great leveler? U.S. EMU JP EM Capital input 1.7 1.5 0.8 5.3 In our growth accounting framework, the key challenge facing Assumed labor offset from automation most developed markets is the decline in labor force growth due +0.2 +0.6 to aging and reduced birthrates. The most extreme example7 is Labor input 0.7 0.5 -0.3 1.8 Japan, where population decline subtracts 0.25% per annum from Labor force growth 0.5 -0.1 -0.4 0.6 our long-term growth estimate (visible in our example calculation Hours 0.0 0.2 -0.1 0.0 in Exhibit 4 as the product of 0.4% assumed shrinkage in the Human capital (skills) 0.2 0.4 0.2 1.3 8 labor force and the 60% labor share in the economy). TFP 0.6 0.5 0.5 1.0 In contrast, labor force growth in the U.S. adds 0.3 percentage Assumed TFP boost from automation 1.0 1.0 1.0 1.0 point (ppt) per annum (again, we assume a 60% labor share in the Labor share 0.6 0.6 0.6 0.5 economy). Simply put, just over half—nearly 0.6%—of the 1.1% Capital share 0.4 0.4 0.4 0.5 differential between our Japan and U.S. GDP growth assumptions Original trend 1.7 1.4 0.6 4.4 is explained by the different trajectories of labor force growth. In Trend with TFP and labor impact 2.7 2.5 1.9 5.4 aggregate, emerging markets are not expected to suffer labor Source: J.P. Morgan Asset Management. force shrinkage over our LTCMA time horizon, but there are wide divergences among specific EM nations—for example, between India’s fast-growing labor force and Russia’s rapidly shrinking one. Automation: The end of the cheap labor advantage for emerging markets? Automation has the potential to narrow these growth differences, What is the impact of automation on emerging economies? Here we offsetting shrinkage in the labor force. Thinking of technology in note that globalization has substantially benefited emerging this way—as a means of making up the drop in labor force from markets, as the outsourcing of manufacturing from developed population aging—would imply narrowing the spread in growth markets has offered a means to accumulate productive capital and rates across developed economies. In our calculation in Exhibit 4, develop a skilled workforce. Relatively cheap EM labor has sparked we simply assume that increased automation exactly offsets the and sustained much of that outsourcing, but if factory automation impact of any negative labor growth numbers,9 thereby shrinking reduces its appeal by reducing developed economies’ domestic resulting GDP growth differences. Clearly, this is somewhat production costs, current and future generations of EM economies speculative, and in reality there is no reason this could not go might have to find new sources of economic growth. Additionally, further. And while this would certainly be a boon to the affected emerging markets are also likely to receive relatively less of a economies, it would also raise the capital share in the economy at productivity boost from automation, given that their still relatively the expense of labor, an issue we discuss in a later section, lower labor costs diminish their incentive to automate. “Challenges for labor and consumers, and a changing role for government?” Overall, then, at least over the LTCMA horizon, we think automation will lead to a leveling of growth differences within both EM and DM economies, but also perhaps a small narrowing of the EM vs. DM growth advantage.

7 Europe is also challenged, but to a much smaller degree over our LTCMA time frame. 8 The impact is also negative in the euro area, but much smaller. 9 Therefore, this only affects Japan and the euro area in our example. Note that given the 60/40 labor-to-capital shares in the economy, the boost to capital growth required to offset a given labor shrinkage is 1.5x larger.

22 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE IMPACT OF TECHNOLOGY ON LONG-TERM POTENTIAL ECONOMIC GROWTH

GROWTH ASSUMPTIONS WHERE IS TECHNOLOGY MOST The more clear-cut impact of automation within our LTCMA growth accounting framework is to boost total factor productivity growth, INVESTIBLE TODAY? analogous to faster technological progress. For simplicity’s sake, in In an investing process founded on fundamental research, the numerical example in Exhibit 4 we represent this effect by just we divide our analysis into key themes that focus on the conservatively adding 1.0%, the bottom end of the range of most dynamics we feel will be important drivers of asset prices over studies of the likely productivity impact from automation. While in the medium term. Among those themes is the widespread this simple example everyone benefits by the same total amount adoption of existing and emerging technologies. We believe from the boost to TFP, clearly a 1.0% gain will feel much more understanding how these key technologies impact industries and the economy may offer the greatest investment potential meaningful in an economy with 0.5% trend growth than in one in the current environment: with 2% trend growth. However, implementing new technologies— such as investing in robots on the production line—might also be • Cloud computing is offering ubiquitous, flexible and expected to appear in capital deepening.10 Further, to compete with on-demand access to a shared pool of computing resources. While delivering substantial cost savings, cloud computing automation, the human jobs of tomorrow will require increased is also fueling a wave of start-ups ready to disrupt education and thus skills deepening. Of these latter two factors, for incumbent companies in the simplicity in our numerical example we only include the impact software industry. from potential labor force replacement into faster capital growth. • The Internet of Things—the connection of non-computer The overall impact on growth from these two aspects is as devices, such as sensors, control systems, white goods and a leveler of differences. Picking the two extreme ends of the DM cars to the internet—will increase connectedness among people, processes and physical things, generating new spectrum, in our example the U.S. economy would go from business models, such as usage-based pricing in the growing almost three times as fast as Japan to growing slightly insurance, telecommunications and energy sectors, as well less than one-and-a-half times as fast. Similarly, the relative as improving inventory control. advantage of emerging markets over developed markets shrinks. • Artificial intelligence will enable machines to perform a Many of the effects described above reflect the potential upside wider range of tasks, from autonomous driving in commercial transportation to computer-assisted diagnosis in to trend growth and paint a picture of an upper bound to our health care to robo-advising in the financial services sector. growth projections if all goes smoothly. However, the timing and extent of any economic boost is difficult to predict, and while • Robotics will result in widespread automation and lead to we’re optimistic that at least some of these gains will be realized, increased productivity and competitiveness. Industrial companies will “re-shore” their manufacturing base, while we are also keen to find tangible investing opportunities today retailers will fully automate the store experience with self- around this theme. In the accompanying box, “Where is checkout and drone deliveries. technology most investible today?” we highlight five areas • Blockchain technology could completely disintermediate the where we believe the early effects of technological change on the settlement and recordkeeping of transactions. Blockchain- world economy are investible today. enabled “smart contracts” will facilitate, verify and enforce the execution of contracts and reduce transaction costs.

10 Increased capital per worker.

J.P. MORGAN ASSET MANAGEMENT 23 TECHNOLOGY, PRODUCTIVITY AND THE LABOR FORCE

CHALLENGES FOR LABOR AND CONSUMERS, AND A All of the above only serves to emphasize the importance of skills CHANGING ROLE FOR GOVERNMENT? deepening through education and retraining (Exhibit 5). In the past, this has allowed workers to benefit from technological Reskilling the labor supply innovation. Further, it has made it possible for economies and the labor force to flourish through successive episodes of disruptive Our growth accounting framework described above estimates technology, from Jethro Tull’s14 seed drill to the internet. The potential growth over long periods by focusing on the supply side of future needn’t be dominated by machines or humans; in an the economy—that is, its productive potential. Embedded in that optimal world, machine and human will productively coexist. Case framework is the implicit assumption that displaced labor is in point: chess players and chess supercomputers. In The Second smoothly redeployed, generally maintaining something close to full Machine Age, McAfee and Brynjolfsson note that although even employment. However, one significant worry surrounding near-term chess grandmasters now have no realistic hope of beating the technological advancement is the reduction in the number of best chess supercomputers, an average chess player in human jobs available in the economy. Such concerns date back at combination with an average chess program can still prevail. least as far as the first industrial revolution. From the automation of Training displaced workers to function in combination with textile manufacturing in the 19th century11 to the recent digitization technology seems an obvious step in skills deepening. But, of music and film,12 new technology has sparked fears of job paradoxically, urging human workers to be more human might destruction—fears that have mainly proved unfounded. also enable them to keep their grip on functions that will probably remain beyond the scope of automation and AI—at least, in the Our base case is that this historical pattern will hold—at least over intermediate term. our 10- to 15-year forecast horizon—and that the labor force will continue to adapt. But this outcome depends crucially on human skills keeping pace with technological advancement. Many past In the past, skills deepening through education and retraining episodes of technological advancement were disruptive to specific allowed workers to benefit from technology, but widespread industries or processes, in particular highly manual or labor- automation across many sectors at once creates a challenge to intensive activities. Reskilling or redeploying labor into other lower skilled workers that is already apparent in labor data functions—or, indeed, new functions explicitly created with the EXHIBIT 5: LABOR FORCE PARTICIPATION BY EDUCATIONAL ATTAINMENT VS. NUMBER OF INDUSTRIAL ROBOTS new technology—generally prevented mass labor displacement Global stock of operational industrial robots (RHS) and eventually boosted low skilled worker productivity. High school or less Some college Bachelor’s or more Prime-age male labor force participation, Worldwide estimated stock of However, the impact of the current wave of automation has so far by educational attainment (%) operational ind. robots (’000s) been skills-biased. That is, it has enhanced the productivity of highly 100 3,000 skilled workers while undercutting the prospects for low skilled 96 2,500 workers. The functions at greatest risk are still likely to be routine 2,000 physical jobs that can be fully automated. These collectively account 92 for 13% of U.S. wages and 18% of time spent in all U.S. 1,500 13 88 occupations —affecting industries such as accommodation/food 1,000 services and manufacturing. Going forward, the data computing and 84 processing powers of emerging technologies will put many routine, 500 non-physical jobs at risk, including white collar administrative 80 0 functions that span industries. The jobs least vulnerable to ’65 ’75 ’85 ’95 ’05 ’15 technological disruption are likely to be non-physical and non- Source: Bureau of Labor Statistics Current Population Survey; Council of Economic routine and generally include functions that require interpersonal Advisers calculations as of June 2016; International Federation of Robotics as of August 2017. skills and expertise—a core human element in many sales, communications, artistic, cultural, health care or management jobs.

11 The Luddite rebellion of 1811-13 was driven by fears that new weaving machines would lead to mass unemployment of textile workers. 12 CD sales fell from almost 800 million in 2000 to just 150 million in 2016 as music streaming took over from conventional album sales (gloriousnoise.com, Nielsen music via Billboard). 13 McKinsey Global Institute, “A future that works: Automation, employment, and 14 Viscount Jethro Tull developed the seed drill in 1701; it is credited as the first major productivity.” automation in agriculture.

24 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE IMPACT OF TECHNOLOGY ON LONG-TERM POTENTIAL ECONOMIC GROWTH

Government will play an important role in providing skills, training The notion that economic demand, especially consumption, will and education, but its track record in this area has been patchy, at keep pace with the potential levels of production that automation least in part because it is difficult to keep educational content might allow cannot be taken for granted if those consumers have relevant in a fast-changing global economy. Given companies’ role fewer jobs or lower salaries. For this reason, we believe that the as the driving force of many of these changes, there is scope for policy debate will focus on two key areas: first, the new roles and innovation around tax structure and other incentives—such as the responsibilities government may need to assume in maintaining accounting treatment of corporate investment in human skills—to the purchasing power of consumers, and by extension the demand motivate companies to promote skills deepening and help side of the economy; and second, incentivizing education and optimize processes that combine human and machine labor. reskilling the broad workforce to rapidly adapt to new technologies and fulfill new job functions. As we explain in the following section, Maintaining consumer demand proactive policies to prepare the labor force for emerging technologies are set to be important in maintaining social and In prior sections of this paper, we have focused on technology’s political stability as automation and AI become more widespread. impact on the supply side of the economy. We now shift our perspective to a consideration of the economy’s demand side. This potential threat to demand is not the only contentious subject for governments, as it relates to a declining labor share of income. One of the main threats from workforce automation is that a Even if the workforce doesn’t face imminent decline—and with greater share of output, all else equal, is likely to accrue to a employment statistics16 showing quite the opposite, this isn’t an concentrated group of capital holders rather than more evenly to immediate risk—the level of real wages and the labor share of the labor as wages. This risks increasing inequality and may weigh on economy are already displaying worrying trends. In particular, real consumer demand (Exhibits 6A and 6B).15 From an individual wage stagnation, growing inequality and a diminished range of job firm’s perspective, it might be profitable to replace a human opportunities appear to be stoking resentment in some quarters. worker with a robot, but that robot will not need the shampoo, The recent rise in populism across developed economies—the vote coffee, mortgage advice and myriad other consumer goods and share for populist fringe political parties is at its highest point services that its displaced human equivalent once did. Simply put, since the 1930s17—points to challenges that a disaffected labor once we look at the economy in aggregate, we must acknowledge force might create. Thus far, anger has been directed mostly at that one sector’s displaced labor may be another sector’s the forces of globalization, especially corporate outsourcing (often disenfranchised consumers. decried as “sending jobs overseas”).

16 Prevailing labor data in the U.S. show the lowest level of unemployment since 2001 and robust demand for labor in most industries. 15 The important subject of income inequality is beyond the scope of this paper. 17 Bridgewater Associates and Lombard Street Research.

The level of real wages and the labor share of the economy are already displaying worrying trends, particularly given that the lowest income brackets spend the greatest share of their income on consumption EXHIBIT 6A: U.S. PRE-TAX INCOME BY INCOME QUINTILE EXHIBIT 6B: MARGINAL PROPENSITY TO CONSUME BY INCOME QUINTILE* (INDEX, 1990=100)

Lowest Second Third Fourth Highest Overall 224 140

130

120 124

Whole 110 93 79 economy 62 100

’90 ’94 ’98 ’02 ’06 ’10 ’14 LOWEST SECOND THIRD FOURTH HIGHEST

Source: Bureau of Labor Statistics Consumer Expenditure Survey; data as of August 2017. * Marginal propensity to consume = average expenditure / pre-tax income.

J.P. MORGAN ASSET MANAGEMENT 25 TECHNOLOGY, PRODUCTIVITY AND THE LABOR FORCE

Productivity gains have historically raised the demand for labor. But the scale and speed of today’s technological changes mean we can’t take a repeat of this outcome for granted EXHIBIT 7: UPSIDE AND DOWNSIDE SCENARIOS OF THE ECONOMIC IMPACT OF AUTOMATION

DEVELOPED MARKETS EMERGING MARKETS

• Automation fills labor force shortfalls, boosting supply side • Automation boost to productivity growth muted relative to potential of developed markets with aging populations developed markets, given lower wages, but impactful in “last • Automation boosts total factor productivity growth mover advantaged” scenarios • Net share of displaced workers is reskilled or subsidized for a time, • Displaced workers magnified given developed markets’ reliance on maintaining purchasing power; wage gap and inequality emerging markets for low skilled tasks, but net share of displaced UPSIDE trends subside workers reskilled or subsidized to maintain purchasing power • Faster absolute growth; spread of growth rates across • Faster absolute growth, but emerging markets’ growth advantage developed markets narrows over developed markets may narrow

• Failure of public policy to react means net share of displaced • Automation boost to productivity growth muted relative to workers lose purchasing power developed markets, given lower wages. Displaced labor impact is • Wage gap and inequality trends magnify, depressing demand and magnified, given developed markets’ reliance on emerging markets risking populist backlash for low skilled tasks, but policy fails to react • Political uncertainty weighs on market confidence, leading to the • Cross-regional wage gap and inequality trends magnify, depressing growth boost from technology undershooting potential demand and risking political instability DOWNSIDE • Demand shortfall negates growth boost and creates structurally • Developed market service providers use improved global depressed interest and inflation rates communications network to fill DM job needs with cheaper EM workers, depriving local EM markets of key higher skilled workers, while also suppressing DM wages and growth potential

Source: Richard Baldwin, The Great Convergence: Information Technology and the New Globalization, Harvard University Press, 2016; J.P. Morgan Asset Management.

We have yet to see a wave of 21st-century Luddism directed at redistribution risks capital flight. We expect governments to avoid automation, but should the object of populist ire shift that way, it both extremes and play a generally positive role, deploying a range will create a significant policy challenge of maximizing the of policies to help ensure that the economic benefits of automation positive impact of technology and minimizing its potentially and AI are widely spread. disruptive effects. The precise policy prescription will ultimately depend on how The history of technological progress shows that productivity gains emerging technologies reshape the labor market. Where have typically raised the demand for labor, not destroyed it. automation augments human labor, the policy imperatives are However, the scale and speed of today’s technological changes likely to be reskilling and retooling labor. But if automation mean we can’t take a repeat of this outcome for granted, and that substitutes for human labor, then the policy imperatives might shift policymakers have a role to play to encourage a continuation of more to redistribution in order to keep capital and labor in balance this trend. Economies operate most efficiently when incentives are and, more crucially, to prevent a drop in demand. aligned and capital and labor are in equilibrium. Extreme policy in In Exhibit 7, we present upside and downside scenarios of the any direction will likely fail—a pure laissez-faire approach risks economic impact of automation. persistent wage pressure and swelling populism, while excessive

26 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE IMPACT OF TECHNOLOGY ON LONG-TERM POTENTIAL ECONOMIC GROWTH

IMPLICATIONS FOR OUR LTCMA FORECASTS

Technological change—especially automation and AI—is likely to that these forces will roughly offset, so we expect equilibrium have profound effects on the global economy over the long term. rates to be little changed. But even then, the 1%-1.5% boost to We are often struck by how rapidly new technology is being potential equity returns could meaningfully increase 60/40 returns developed and adopted. The early effects—at least over the 10- to as automation and AI become more widespread in the next 10 to 15-year forecast horizon of the LTCMAs—will probably accrue 15 years. slowly. Nevertheless, the groundwork for more substantial changes to the economy will be laid. If automation and AI provide This is, of course, a rosy view, but it does set a reasonable upper the kind of productivity boost suggested by a range of studies, bound to the impact of technology on the economy and asset including those from PwC and the McKinsey Global Institute, of returns over our forecast horizon. By contrast, should automation 1%-1.5%, and technology also essentially offsets population and AI start to displace labor and reduce wage growth, it would decline in some developed nations, then annualized DM growth potentially offset some of the boost from productivity—in turn, and equity returns could be more than a full percentage point weighing on nominal growth forecasts and equilibrium interest higher than currently assumed. Moreover, the dispersion in equity rates, and bringing equity returns back down toward, or even return between developed and emerging markets has scope to below, our base case estimates. narrow meaningfully. Ultimately, modeling the impact of technology on productivity, the The effect on equilibrium interest rates, however, may be more labor force, the economy and government policy is an issue of nuanced. While higher productivity and better growth might extraordinary complexity. Paradoxically, we might need to wait for increase equilibrium yields, at the same time reduced labor a sufficiently advanced level of artificial intelligence to truly bargaining power would likely keep inflation in check, which could understand it. reduce equilibrium yields. On balance, our baseline assumption is

J.P. MORGAN ASSET MANAGEMENT 27 THE IMPACT OF GLOBAL AGING

How demographic change will affect savings, growth and interest rates Benjamin Mandel, Ph.D., Global Strategist, Multi-Asset Solutions Thushka Maharaj, DPhil, CFA, Global Strategist, Multi-Asset Solutions Akira Kunikyo, Client Portfolio Manager, Multi-Asset Solutions Diego Gilsanz, Associate, Multi-Asset Solutions Nika Mosenthal, Analyst, Multi-Asset Solutions

IN BRIEF • The world economy stands on the brink of a sizable, long-term decline in household savings, driven by a global aging process that is set to become faster and more synchronous in the coming decades. In this paper, we consider whether this shift will reverse the global savings “glut” of the early 2000s and apply upward pressure to interest rates.

• Our analysis of the U.S., Germany, Japan and China suggests that an aging-driven decline in savings of 2.6% of GDP will unfold over the next 30 years, translating into real interest rates that are 25 to 50 basis points (bps) higher than they would have otherwise been. In contrast, our Long-Term Capital Market Assumptions (LTCMAs) embed a population-aging effect (through slower labor force growth) that lowers trend growth and rates by ~50bps over the next 10 to 15 years.

• Which force prevails—whether demographics drive rates higher or lower—will depend on how other parts of the economy respond. For instance, whether investment demand wilts under lower trend growth or is buttressed by a transition to a more capital-intensive economy will be a key determinant.

28 LONG-TERM CAPITAL MARKET ASSUMPTIONS HOW DEMOGRAPHIC CHANGE WILL AFFECT SAVINGS, GROWTH AND INTEREST RATES

GLOBAL AGING AND THE IMPACT OF DEMOGRAPHIC CHANGE

Deeply enmeshed in our capital market assumptions is the idea The first reason demographics matter in this context is that the that demographic change influences long-term asset returns. global aging process will become faster and more synchronous in Our interest rate projections are informed by an assumption about the coming decades. The dependency ratio—defined here as the trend growth, itself a function of labor productivity and the ratio of the older to the younger population—has been rising for demographic factors shaping the size of the workforce. A some time in Japan and Germany and is beginning to creep up in reasonable rule of thumb suggests that for each percentage point the U.S. and China.2 According to United Nations population (ppt) change in trend growth, the real interest rate of the forecasts, the overall dependency ratio will rise by 10ppt per economy changes by about the same amount.1 decade over the next 30 years (Exhibit 1).

However, there are important exceptions. One exception arises when the supply of global savings changes so dramatically and The global aging process will become faster and more synchronous in the coming decades quickly that it causes an imbalance in the world’s desired (i.e., ex ante) level of savings vs. its desired level of investment—a EXHIBIT 1: DEPENDENCY RATIOS, 1990-2050 U.S. Germany Japan China Total so-called savings glut. Former Federal Reserve governor and % chairman Ben Bernanke popularized this idea in the mid-2000s to 80 explain why real interest rates were so low and why the U.S. was running such a large current account deficit (Bernanke 2005, 60 2007). He argued that the rapid buildup of savings in emerging market (EM) economies after a spate of financial crises in the 1990s, a sharp rise in the price of oil in the early 2000s and an 40 upswing in household and corporate savings in China caused a shift outward in the global savings supply curve. (That is, 20 it led to a greater supply of global savings at any level of the interest rate.) These increases in the supply of savings, in turn, 0 could only be absorbed by global investment demand once ’90 ’10 ’30 ’50 interest rates moved lower and developed market (DM) economies Source: United Nations, J.P. Morgan Asset Management Multi-Asset Solutions. became net borrowers. Importantly, real interest rates fell by more in the mid-2000s than did estimates of trend growth. The second reason demographics matter is because the Today we stand on the brink of another massive swing in global relationship between labor force growth and interest rates will savings, driven this time by the aging of the global population. arguably be more uncertain going forward. Demographic forces Older cohorts transitioning out of the workforce and into will likely push interest rates in opposite directions. Slower labor retirement will save less as they draw down assets to finance force growth, dragging on trend growth, will put downward consumption. The process is well underway in DM economies and pressure on rates; at the same time, lower saving rates in will hasten in the coming years as U.S. retirement rates continue to retirement will apply upward pressure on rates. More specifically: rise and, eventually, as China’s huge population ages and lowers its On the one hand, demographics will continue to exert downward saving rate. In this paper, we explore the idea that this decline in pressure on trend growth of ~50bps over the next 10 to 15 years, savings will reverse the behavior that accompanied the savings glut reducing interest rates by about the same amount. On the other, of the early 2000s and will apply upward pressure to global real we estimate that the corresponding reduction in savings—all else interest rates. equal—will raise interest rates by 25bps to 50bps over the next 30 years. We are fairly confident that the demographic forces driving trend growth and rates lower will prove to be the more powerful influence in the long run. But we recognize that demographics can buffet interest rates along the way, subjecting our yield estimates to additional uncertainty. 1 This rule is also rooted in economic theory. In neoclassical growth models, the saving rate that maximizes the level of consumption, known as the “golden rule” saving rate, implies that the risk-free rate in the economy equals its trend growth rate. See Solow, Robert M. (1956): “A contribution to the theory of economic 2 The analysis herein will focus on these four economies, which together compose growth.” The Quarterly Journal of Economics 70:1, 65-94. 26% of the global population, 50% of global GDP and 55% of global gross savings.

J.P. MORGAN ASSET MANAGEMENT 29 THE IMPACT OF GLOBAL AGING

LIFE CYCLE SPENDING PATTERNS AND THE FUTURE OF GLOBAL SAVINGS

In defining a reversal of the savings glut, we first relate the large changes in income and consumption between working age and expected swing in the age composition of the population to retirement, we are able to estimate the average saving rate in aggregate savings. We focus on the household saving rate as the each cohort.5 For example, the official gross household saving rate most direct channel and use a simple technique to parse the in the U.S. was 7% of disposable income in 2016, from which we overall saving rate of each country into a saving rate for the estimate a 13% average saving rate for younger cohorts and a younger cohort (under 65) and the older cohort (over 65).3 The -48% average rate for older people. technique is based on the idea of consumption smoothing over the life cycle, since people entering retirement face a large decline in Armed with age-specific saving rates and a forecast for dependency their income but use their accumulated savings to finance ratios, we estimate the effect of aging on average saving rates in consumption.4 Therefore, under our assumptions about average the coming decades. Our estimates are shown in Exhibit 2, with each economy’s saving rate falling roughly in line with the contour of the rise in its dependency ratio. In the U.S., where saving rates 3 We use slightly different definitions of older cohorts across countries, reflecting differences in retirement ages. The cutoff is assumed to be 65 years old in the U.S., have been oscillating around 9% in recent decades, the retirement Germany and Japan. In Japan, retirement age varies by pension scheme, but under of the baby boom cohort (Americans born 1946-64) will subtract current legislation the varying ages will center around 65 in the coming years. The Chinese cutoff is set at 60, the current retirement age for men (the female age is about 2ppt between now and 2030 before stabilizing. In Japan and 55). Germany, where saving rates are generally higher and dependency 4 Another important driver of household savings operates via real estate. In previous ratios have already been on the rise, the swing will be of similar research using the U.S. Survey of Consumer Finances (https://am.jpmorgan.com/ gi/getdoc/1383246462222), we found that residential property alone contributed magnitude between now and 2030. But while U.S. saving rates are 59% of the average baby boomer’s total asset growth since 1989. Unfortunately, a similar forensic decomposition of household balance sheets by age group is not possible for the other countries, given data limitations, and that is why we focus on 5 Our calculations assume a decline in income of 50% and a decline of consumption the flow of income diverted to savings as opposed to the stock of savings. of 15% in old age.

Life cycle saving patterns will exert downward pressure on household saving rates EXHIBIT 2: HOUSEHOLD SAVING RATES, U.S., GERMANY, JAPAN, CHINA, 1990-2050 (%) U.S. GERMANY Ocial Demographic eect Ocial Demographic eect % % 10 20

8 16

6 12

4 8

2 4

0 0 ’90 ’00 ’10 ’20 ’30 ’40 ’50 ’90 ’00 ’10 ’20 ’30 ’40 ’50

JAPAN CHINA Ocial Demographic eect Ocial Demographic eect % % 25 50

20 40

15 30

10 20

5 10

0 0 ’90 ’00 ’10 ’20 ’30 ’40 ’50 ’90 ’00 ’10 ’20 ’30 ’40 ’50

Source: National statistical offices, United Nations, J.P. Morgan Asset Management Multi-Asset Solutions. Note: Throughout this analysis, we use gross household saving rates rather than net rates, given data availability constraints for China. Gross household saving rates are generally higher levels than net rates since they include depreciation and other forms of household capital consumption, but have little effect on the changes in saving rates that we forecast.

30 LONG-TERM CAPITAL MARKET ASSUMPTIONS HOW DEMOGRAPHIC CHANGE WILL AFFECT SAVINGS, GROWTH AND INTEREST RATES projected to stabilize in 2030, in Japan and Germany they are saving rate going forward, given that the age-specific rates expected to continue falling, bringing the total decline in the saving themselves might be driven by changing demographic factors. rate to 4%-5% through 2050. Indeed, according to our calculations, if China’s working age saving rate were to revert to where it was at the beginning of the In China, where the puzzlingly high level of household savings has 2000s, its overall saving rate would be closer to 25% by 2050. It been the subject of a voluminous literature (see, for instance, could also be the case that these types of forecast errors are Modigliani and Cao 2004 or Ma and Yi 2010), demographic offsetting across countries: Germany, for example, has had a effects are expected to bring down saving rates steadily, from rising dependency ratio since 2000 without any notable decline 38% at present to 31% in 2050. China’s case, in particular, allows in its saving rate. We discuss Japan’s experience in the us to examine whether it is reasonable to assume a constant accompanying sidebar, “The Case of Japan.”

THE CASE OF JAPAN Our analysis makes a series of assumptions about the nature of to that, modest tax incentives for saving had been scaled back over savings in the wake of a demographic shift. For example, we’ve the course of the 1980s. Delayed retirement is an example of a embedded an assumption that households smooth out consumption factor that would have slowed the decline in saving rates, though over their life cycles and therefore have lower saving rates older worker labor force participation has actually fallen slightly in retirement. We have also focused on the household sector, since the mid-1990s. paying little attention to the effects of demographic transition on We also note that, even as household saving rates continued corporate or government savings. Japan’s experience over the past to decline in the mid-1990s, the contribution of corporates to quarter century provides a case study in addressing whether these the country’s gross national savings (i.e., the total including assumptions are reasonable. household, corporate and government savings) was steady at To put the Japan case in context, the economy’s dependency ratio ~4%-5% of GDP (Exhibit 4). The fact that corporate savings swung has nearly tripled since the early 1990s. Consistent with our thesis from negative to positive raises the possibility that lower trend that a higher share of older age cohorts translates into lower growth—largely driven by demographics—reduced incentives for savings, household saving rates have fallen dramatically over Japanese companies to borrow. However, that swing may have that period, from 19% in 1995 to 7% in 2015 (Exhibit 3). Using simply reflected the deflation of the 1980s bubble. The lack of our estimates of age-specific saving rates from the mid-1990s, any discernible correlation between weakening demographics demographic changes would have accounted for 4.9ppt (40%) of and steady corporate savings after the initial change in the 1990s the total 12.7ppt decline in the saving rate. Of course, demographic is consistent with the idea that the country’s financial crisis change was but one factor that caused saving rates to decline. was a more powerful influence on corporate savings. Overall, Others include the gradual recovery of household balance sheets these patterns leave us fairly comfortable with our focus on following the financial crisis in 1990. And, even prior demographically driven shifts in household savings.

Japanese household saving rates have fallen dramatically since In Japan, corporate contributions to gross national savings have the mid-1990s, with demographics accounting for about 40% of held relatively steady as household saving rates have declined the decline EXHIBIT 3: HOUSEHOLD SAVING IN JAPAN EXHIBIT 4: CONTRIBUTIONS TO TOTAL SAVING RATES IN JAPAN

O cial Demographic e ect Government Corporation Household 25 % 15 20 10

15 5 12.7% 4.9% 0 10 -5 5 -10

0 -15 ’90 ’95 ’00 ’05 ’10 ’15 ’80 ’85 ’90 ’95 ’00 ’05 ’10 ’15

Source: Bank of Japan, Cabinet Office of Japan, Haver Analytics, J.P. Morgan Asset Management Multi-Asset Solutions; data through July 2017.

J.P. MORGAN ASSET MANAGEMENT 31 THE IMPACT OF GLOBAL AGING

THE EFFECT ON REAL INTEREST RATES

Having established a connection between demographic transition (i.e., the price elasticity of savings and investment, respectively).6 and savings, we now link our projections of slowing household Aggregating across our savings projections for the U.S., Germany, savings to real interest rates. A stylized way to render that link is Japan and China, we estimate that aging will subtract 1.2ppt of through the supply and demand for global savings. The idea is GDP over the next 10 years and 2.6ppt cumulatively over the next that the aging of the global population— as measured here in the 30 years; these are the leftward shifts in the supply curve shown U.S., Germany, Japan and China—represents an inward shift in the in Exhibit 5.7 Translating these shifts into increases in the real savings supply curve (Exhibit 5). If demographic forces influence interest rate, aging will push up rates by 15bps over the next 10 interest rates primarily through supply—and not by shifting the years and 35bps over the next 30 years. Allowing for some demand curve for investment—the contraction in household uncertainty about what the true elasticities are, the effect is in the savings we’ve identified will put upward pressure on the price of range of 25bps-50bps over the next 30 years. savings, which is the real interest rate. 6 Our baseline elasticity assumptions are based on work by Rachel and Smith Given that demographic change is an increasingly global influence (2015). In their work, βs=0.5 in a savings supply equation of the form: s ln(r)=αs+(1/βs)ln(y ) and βi=-0.7 in an investment demand equation of the form: on savings, we model the corresponding shifts in global supply s ln(r)=αi+ln(1/βi)ln(y ). We present results for a range of elasticities: βs=[0.1,0.9] and demand. We believe that this mechanism will become and βi=[-0.3,-1.1]. increasingly influential over time. For instance, we expect the Our starting point for the current global real interest rate is something we would consider to be a “normal” rate in the absence of quantitative easing and other effect to be larger in the coming decades than it was in Japan, factors pushing around the net supply of safe assets. To estimate what that normal which was at the forefront of developed market demographic real rate would be, we use an assumption similar to the LTCMA framework for fixed income, linking it to trend real growth. Since we think that trend global growth shifts in the 1990s. In that case, the savings effects were more is currently ~3%-3.5%, that is the starting point for the global real interest rate local in nature and domestic interest rates continued their in Exhibit 5. The starting point for the national savings as a share of GDP (25%) is a weighted average across countries and comes from the IMF World Economic downward trend. Outlook. 7 These projections are arguably conservative compared with what would be We calibrate the effect on real interest rates using our estimates obtained using prior estimates of the link between dependency ratios and overall of the aging-driven decline in savings in the coming decades, as savings. Rachel and Smith (2015) present cross-country evidence that a 1ppt increase in the dependency ratio tends to lower savings by about 50bps of GDP, well as estimates of the slope of the supply and demand curves implying a decline in savings of 4%-5% of (overall) GDP over our forecast horizon.

If demographic forces influence interest rates primarily through supply, a contraction in household savings will put upward pressure on real interest rates EXHIBIT 5: SUPPLY AND DEMAND FOR GLOBAL SAVINGS (%) 6.0

Next 10 years Next 30 years 5.0

4.0

3.0 Supply of Investment World real interest rate (%) savings demand

2.0 15 20 25 30 35 World saving rate (% GDP)

Source: Rachel, Lukasz and Thomas D. Smith (2015): “Secular drivers of the global real interest rate.” Bank of England Staff Working Paper No. 571. J.P. Morgan Asset Management Multi-Asset Solutions; data as of July 2017.

32 LONG-TERM CAPITAL MARKET ASSUMPTIONS HOW DEMOGRAPHIC CHANGE WILL AFFECT SAVINGS, GROWTH AND INTEREST RATES

CONCLUDING THOUGHTS

As we have discussed, a tension between two opposing forces While lower trend growth will depress investment, all else equal, will play out in the way that aging populations will influence we also see several factors boosting investment in the coming interest rates. As more workers retire and workforce growth years. Take business investment spending or capital expenditure slows, the demographic change will act as a source of downward (capex), for example. Notwithstanding the fact that capex trends pressure on both economic growth and equilibrium interest generally follow those of overall growth, we expect recent cyclical rates. However, the change will also coincide with a large distortions in capex to normalize and for investment in technology downshift in savings as those workers save less in retirement, (and intellectual property more generally) to support overall and in this way it will apply upward pressure to rates. investment. Perhaps the biggest question mark with regard to aging and investment is whether the demographic process we Which force will prevail over our 10- to 15-year forecast horizon? It have described here spurs additional developments in labor-saving will depend in part on how other areas of the economy respond. technologies and automation, which would cause the economy to Viewed through the lens of our supply and demand framework, to become more capital intensive and support investment spending get to the lower equilibrium interest rates that theory implies along the way. even as the supply curve of savings is shifting inward, the rest of the economy would need to increase savings somehow, or All in all, these crosscurrents make it difficult to predict whether investment demand would need to shift inward as well. we will see a strong reversal of the global savings glut and a corresponding upswing in interest rates. As a case in point, in the As households lower their saving rates, aging populations may 1990s and 2000s Japanese government bond yields did not show well boost savings in the rest of the economy. Governments may any discernible upward impulse as household savings declined, be required to shore up savings to meet future entitlement perhaps swamped by the effects of weak trend growth or the obligations, and businesses may take on less debt as they international drivers of interest rates. What is clear, though, is that anticipate a slower-growth future. It is also possible that China’s the accelerating and increasingly globally synchronized nature of growing weight in the world economy will pull global saving rates demographic change will put this hypothesis to the test in the closer to China’s own astronomical levels. In other words, the fact coming decades. And while the arrival of higher interest rates is that China is growing relatively quickly means that its contribution by no means an inevitable outcome of the demographic shifts to to global savings flows will actually grow, not shrink. Finally, from come, we view it as an important upside risk. the U.S. perspective, to the extent that foreign savings fall by more than that in the U.S., the U.S. current account deficit would likely shrink.

References 1 Bernanke, Ben S. (2005): “The global saving glut and the U.S. current account deficit.” Remarks presented at the Sandridge Lecture, Virginia Association of Economists, Richmond, Virginia. https://www.federalreserve.gov/boarddocs/speeches/2005/200503102/#table1 2 Bernanke, Ben S. (2007): “Global imbalances: Recent developments and prospects.” Remarks presented at the Bundesbank Lecture, Berlin, Germany. https://www.federalreserve.gov/newsevents/speech/bernanke20070911a.htm 3 Ma, Guonan and Wang Yi (2010): “China’s high savings rate: Myth and reality.” BIS Working Paper #312. 4 Modigliani, F. and S.L. Cao (2004): “The Chinese saving puzzle and the life-cycle hypothesis.” Journal of Economic Literature, 145-70.

J.P. MORGAN ASSET MANAGEMENT 33 PENSION INVESTMENT STRATEGY

Matching cash flows and managing liquidity in maturing pension funds Sorca Kelly-Scholte, FIA, Global Strategist, Global Pension Solutions Timothy Lintern, Analyst, Global Strategist, Multi-Asset Solutions

IN BRIEF • Many maturing developed market pension plans are becoming steadily more cash flow negative. The challenge is most pronounced for closed or frozen plans, whose payments to beneficiaries are increasing and contributions to fund younger participants’ benefits are declining.

• The combination of underfunding, negative cash flow and a low return environment can lead to a downward spiral in which funded status falls irretrievably (without additional contributions).

• Rising rates alone are unlikely to stop the fall, particularly in the U.S. Applying our 2018 Long-Term Capital Market Assumptions (LTCMAs), we demonstrate that most of our projected rise in U.S. yields is already incorporated in liability discount rates through current market forward yield expectations. Closing the remaining gap is unlikely to be enough to provide the required uplift for U.S. pension funding levels. The potential implications for European plans are more encouraging, although rising rates alone are still not expected to solve the funding shortfall.

• For underfunded pension plans with substantial allocations to risk assets, liability-driven investment (LDI) strategies are only partially addressing near-term cash flow servicing—and those using the derivatives market to implement liability hedging may be compounding the problem, especially in a rising rate environment.

• We expect that chronically negative cash flows will lead corporate pensions to shift toward buy and maintain1-style credit portfolios and income-generative assets, structured to manage both short-term liquidity requirements and duration risk. We expect that this will sustain the pension fund demand for bonds through our 10- to 15-year assumptions period.

1 Buy and maintain portfolios are constructed based on an investor’s funding profile; bonds are selected based on the issuer’s ability to service its debt until maturity and are intended to be held to maturity unless the probability of default for a particular bond held rises to unacceptable levels (in which case, it is sold).

34 LONG-TERM CAPITAL MARKET ASSUMPTIONS MATCHING CASH FLOWS AND MANAGING LIQUIDITY IN MATURING PENSION FUNDS

INTRODUCTION also apparent in the UK and Europe, although data prior to 2014 is less reliable, given the small number of plans represented. Net Many of the developed world’s maturing pension plans are now cash flow positions were briefly positive in the post-crisis years, reaching a tipping point as they move toward final runoff. They find largely as a result of crisis deficit repair contributions made in themselves in a negative cash flow position in which they are paying response to plummeting funding levels at that time. Contribution out more in benefits than they are receiving in contributions. The patterns tend to be lumpy under U.S. funding regulations, which need to service cash outflow is an additional burden for investment can cause the distribution of cash flow positions across plans to be portfolios already straining to achieve long-term investment goals, quite wide in any given year. The underlying trend in the data at a burden that will weigh ever more heavily as plans mature. the median level is clear, however, and our analysis of the effect of We examine how negative cash flow impacts funding, risk and removing outliers (spikes in contributions or benefit outflow, return for pension plans and provide insight on how plans are where the amount in a given year is substantially greater than the likely to adapt their investment strategies in response, average over the full period) confirms this. Today almost three- taking into account current capital market conditions and quarters of corporate defined benefit plans have negative cash our 2018 Long-Term Capital Market Assumptions. We expect flows, and we expect the trend to affect more and more corporate sustained demand from pension funds for high quality bonds and funds as they mature. income-generative assets over the 10- to 15-year horizon of our Indeed, there are good reasons to believe that this negative cash LTCMAs. This will be associated with a shift away from hedging flow trend might accelerate. The majority of corporate pension duration with derivatives and toward matching cash flows with plans in the UK and large plans in the U.S. are now closed to new physical assets—in particular, buy and maintain credit and core participants, and many are frozen—closed to any further accrual real assets. Once maturing plans are more fully funded and of benefits. Once a fund is closed or frozen, the march toward a matched, they will continue to be big holders of bonds, but they negative cash flow position is inexorable because offsetting will have little need to buy further bonds through runoff. Beyond contributions to fund younger participants’ benefits decline, most our LTCMA time frame, therefore, we envisage a drop-off in rapidly for those funds that are frozen. demand from defined benefit pension funds, with few obvious candidates to replace that demand. More immediately, contributions continue to be elevated in the ongoing bid to shore up funding levels, and it is these deficit repair contributions that are holding many pensions, in Europe in THE JOURNEY TOWARD FINAL RUNOFF particular, in a cash flow-positive position for now. In the U.S., as Taking the simplest possible definition of net cash flow position— noted previously, contributions at the individual plan level can be total contributions less total benefit payments—Exhibit 1 shows lumpier, but in aggregate contributions are elevated while funding how the net cash flow position has evolved for corporate pension levels are low. Once funding levels reach required regulatory plans in the U.S., UK and Europe. The trend is clear—pension funds standards, however, it is reasonable to expect that contributions will have become steadily more cash flow negative. This is evident in drop back to the levels required to cover the ongoing accrual of the the U.S., where we have the longest, most robust data set. It is dwindling active participants’ benefits.

Net cash flow position is increasingly negative for developed market pension systems EXHIBIT 1: CORPORATE DEFINED BENEFIT NET CASH FLOW POSITION AS % OF ASSETS

% of assets 15.0 U.S. UK Europe

10.0 PERCENTILES 5.0 95th

0.0 75th Mean Median th -5.0 25 5th -10.0 2007 2010 2013 2016 2014 2015 2016 2014 2015 2016

Source: HOLT® (based on accounting disclosures for S&P 500, FTSE 350 and MSCI Europe), J.P. Morgan Asset Management; data as of June 30, 2017.

J.P. MORGAN ASSET MANAGEMENT 35 PENSION INVESTMENT STRATEGY

We expect the deficit repair process to proceed at a slow, steady from the pension system to the insurance system and therefore do pace rather than through sudden, rapid improvements. A sizable not result in a change in aggregate pension cash flow. However, if part of the yield increase embedded in our Long-Term Capital the risk transfer is financed in part by a one-off contribution to Market Assumptions has been priced into bond markets in the U.S. make good any shortfall relative to the insurance contract’s price, as bond yields have risen over the last year, with pension funds the cash flow pattern is changed. What may have been a positive enjoying an associated uplift in funding levels. Consequently, if our net cash flow position for some time is converted to a large one- yield expectations are borne out, most of the future funding level off positive cash flow before the negative cash flow runoff begins. repair in the U.S. will need to come from asset returns and contributions rather than from rising interest rates. In a low return world, this repair process will be slow. In the UK and Europe, by PUBLIC PENSION PLANS: ON A SLOWER-MOVING contrast, yields have not risen by as much over the last year and TREND yield normalization in line with our long-term capital market Public pension plans are more likely to be open than corporate expectations has the potential to help lift pension funds out of their pension plans, but they are not immune to negative cash flow current funding stress over the next 10 to 15 years. However, we positions. (Neither are defined contribution [DC] plans immune, as expect this to occur slowly as yields grind, rather than spike, discussed in the sidebar on DC plans and demographic trends.) An upward; we do not expect interest rate movements alone to solve open, public defined benefit plan, depending on its funding the pension funding challenge (Exhibit 2). arrangements and demographic profile, can become cash flow negative over time—but this is likely to happen more gradually than CHANGING THE SHAPE OF CASH FLOWS for a closed or frozen plan. There are several forces at work with the potential to speed up the The U.S. public pension system is one example: It is substantially journey to final runoff or, at least, change its route. On the cash flow negative and has been so for the last decade. The UK’s regulatory front, for example, the “freedom and choice” agenda in Local Government Pension Scheme (LGPS) is similarly in a the UK—reform that gives people greater access to their pensions—is negative cash flow position, but to a much lesser degree (Exhibit currently driving substantial activity in the form of individual lump 3). The trends for these largely open defined benefit systems may sum transfers out of defined benefit pension plans, a trend that may not be as inexorable as for closed plans. However, aging accelerate negative cash flows and runoff. demographics can push an open fund slowly but surely toward negative cash flow. The Public Service Pension Fund of Taiwan is a Risk transfer to insurance companies by means of a buy-in case in point—from 1996 to 2015, its net cash flow declined from or buy-out contract is another avenue for changing cash flows. 97% to -2% of assets (Exhibit 4). Strictly speaking, these contracts involve a transfer of cash flows

Rising rates alone are unlikely to solve the pension funding dilemma—particularly in the U.S. EXHIBIT 2: POTENTIAL UPLIFT IN CORPORATE PENSION FUNDING LEVELS DUE TO INTEREST RATE NORMALIZATION

U.S. UK Netherlands** Current funding level (%) 84 89 106 Current average duration (years) 12 18 21 Estimated duration range in 2028 (years) 6–8 12–14 18–20 Estimated hedge ratio* (%) 50 50 39 Potential uplift in funding levels due to gap between current market forward yield 1–2 3–7 6–11 expectations and LTCMA expectations (ppt) Source: De Nederlandsche Bank, Milliman, Pension Protection Fund PPF 7800 Index, J.P. Morgan Asset Management; data as of June 30, 2017. Yield data from Bloomberg as of July 19, 2017. *Hedge ratio measures the sensitivity of plan assets to a change in interest rates relative to the sensitivity of plan liabilities to the same rate change. A 50% hedge ratio implies, for example, that for a given interest rate increase (decrease), assets would decline (increase) half as much as liabilities. **Note that Dutch pension funds must target at least 110% funding before any indexation can be granted, and build reserves of 123%, on average (scheme dependent), before full indexation can be paid to members.

36 LONG-TERM CAPITAL MARKET ASSUMPTIONS MATCHING CASH FLOWS AND MANAGING LIQUIDITY IN MATURING PENSION FUNDS

Pension systems with largely open plans can still be in a negative Aging demographics can nudge even an open plan toward a negative cash flow position cash flow position EXHIBIT 3: PUBLIC PENSION PLAN NET CASH FLOW POSITION EXHIBIT 4: PUBLIC SERVICE PENSION FUND OF TAIWAN CASH FLOWS TAIWAN NEW DOLLAR 100 MILLION % of assets Income Outgo Net cash ﬂow as % assets (RHS) 4.0 U.S. UK NT$100mn 800 100 2.0 PERCENTILES 80 0.0 600 th 95 60 75th -2.0 Median th Mean 25 400 40 -4.0 5th 20 -6.0 200 0 -8.0 0 -20 -10.0 ’11 ’12 ’15 ’13 ’97 ’16 ’14 ’10 ’98 ’01 ’99 ’96 ’07 ’02 ’05 ’03 ’08 ’09 ’06

’07 ’10 ’13 ’16 ’16 ’04 ’00

Source: Department for Communities and Local Government, Public Plans Data (www. Source: Public Service Pension Fund, J.P. Morgan Asset Management; data as of July publicplansdata.org), J.P. Morgan Asset Management; data as of June 30, 2017. U.S. data 19, 2017. based on actuarial (smoothed) value of assets.

Nonetheless, the trend will be much slower for open plans; in the FORCED SELLING: Negative cash flow also puts the fund at risk UK and the U.S., it is more likely that an upward shift in contribution of becoming a forced seller of assets to service cash flows. If this levels will offset any underlying demographic trend, at least for the occurs after periods of market drawdown, a loss is locked in. Of time being. Corporate defined benefit plans have, in many regions, course, the symmetrical argument on the upside means that gains been obliged to very rapidly recognize reduced future return may be locked in. The overall effect of a negative vs. positive cash expectations via discount rates linked to interest rates, and to make flow position is to amplify rather than smooth funding level corrective contributions accordingly. By contrast, public pension volatility. For many corporations with large legacy pension funds, funds have been much slower to lower their long-term return this additional funding level volatility is unlikely to be tolerable. expectations. As return expectations are gradually revised downward, consistent with the mostly moderate growth, low return THE LIQUIDITY SQUEEZE: A fund in negative cash flow will be at world described by our 2018 LTCMAs, contributions will eventually increased risk of liquidity squeezes and will need to address the need to be adjusted upward (or benefits adjusted downward). operational challenge of regular divestment of assets. At modest levels of negative cash flow, such as those seen in the UK’s LGPS, these challenges can be readily addressed by tilting toward FUNDING AND INVESTMENT CHALLENGES income strategies. For plans with higher levels of negative cash All of this—underfunded plans, negative cash flows, a low return flow, current levels of yield on low risk financial assets are not environment and the inability of rising rates alone to shore up sufficient to deliver the required level of cash flow, and declining pension funding—matters because the combination creates risk tolerance will limit the reach into higher risk assets for yield. additional challenges for both the funding and the investment of For these funds, which include a large portion of Western defined benefit plans. corporate defined benefit plans, deeper structural changes to portfolios will be required. THE “SINKING STONE” EFFECT: If a plan is underfunded, then having more cash flowing out than coming in acts as a further drag MANAGING DURATION AND LIQUIDITY RISK: To avoid the need on the funding level—rather like trying to fill a leaking bucket. All to regularly divest assets, cash flows can be serviced with else being equal, this negative cash flow will raise the level of return principal payments from high grade bonds as well as general required to achieve funding targets. If returns are insufficient, investment income. Although corporate pension funds have been funded status will drop further. The more deeply negative the cash de-risking, liability-driven investment portfolios have been flow and the more underfunded the pension plan becomes, the structured principally to manage duration risk across the full term heavier the drag. At a certain point, investment returns alone will structure of the liabilities rather than to manage near-term be unable to stabilize funded status; it will keep falling irretrievably, liquidity requirements. If a pension plan is fully funded and like a sinking stone, unless contributions are injected. invested entirely in bonds, the bond portfolio can be structured to

J.P. MORGAN ASSET MANAGEMENT 37 PENSION INVESTMENT STRATEGY

DEFINED CONTRIBUTION SYSTEMS: SUBJECT TO THE SAME DEMOGRAPHIC TRENDS

Defined contribution (DC) retirement savings systems are, in Demographic trends are a major driver of defined contribution general, much less mature than defined benefit systems, but in system aggregate net cash flows time they too will start to decumulate. In the U.S., the corporate U.S. AND AUSTRALIAN DEFINED CONTRIBUTION SYSTEMS—NET CASH defined contribution system has been in negative cash flow for FLOW AS % OF ASSETS the last few years. This may be due in part to transfers from Australia U.S. DC plans into individual retirement accounts (IRAs)—which % 16.0 would not contribute to net dissaving from the overall pension market. If transfers to IRAs are part of the explanation, the 12.0 Department of Labor (DOL) Fiduciary Rule may slow this DC outflow by focusing greater scrutiny on financial advisors’ 8.0 recommendations to clients to roll over from DC plans to IRAs.* Australia’s DC system remains in positive net cash flow, but 4.0 the downward trend is clear (see exhibit, right). Any declining trend in net cash flow in these open systems will be driven by 0.0 demographics, eventually reaching the point where the retired cohort is drawing income out faster than the younger cohorts are -4.0 paying in.** ’05 ’06 ’07 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16

Source: Australian Prudential Regulation Authority, Cerulli, J.P. Morgan Asset Management; data as of July 19, 2017.

* The DOL Fiduciary Rule holds financial advisors to a fiduciary (vs. suitability) standard—and will generally require advisors to act in the client’s best interest, mitigate conflicts and disclose all fees and commissions. The rule’s definition of fiduciary investment advice includes that related to distributions from, or rollovers to, a plan or IRA. ** For more on the impact of demographics on savings, see “How demographic change will affect savings, growth and interest rates,” J.P. Morgan Asset Management, 2018 Long-Term Capital Market Assumptions.

meet both the duration-hedging and the cash flow-matching In the meantime, and for the longer term, high yielding, stable requirements of the plan. But for underfunded pension plans with cash flow assets are increasingly attractive for pension funds substantial allocations to risk assets, LDI portfolios are only partially seeking to service cash flow requirements, especially if these addressing near-term cash flow requirements. For plans that assets offer the potential for long-term capital growth. This achieve the duration hedge with leverage via the derivatives underscores the merits of real assets for pension investors—in markets, as is commonplace in Europe, increasingly onerous particular, core real estate and infrastructure. We also expect that collateralization requirements in a rising rate environment have the the alternative credit markets will continue to be a key area of potential to compound the cash flow challenge. interest for pension funds seeking stable yield.

RESPONDING TO THE CHALLENGE CONSIDERATIONS BEYOND THE LONG TERM We expect that, over time, chronically negative cash flows will lead What could all this mean for capital markets—beyond the time corporate pensions to shift toward buy and maintain-style credit frame of this year’s Long-Term Capital Market Assumptions? portfolios, structured to manage short-term liquidity requirements In concluding, we share some thoughts on the potential direction as well as duration risk. In the case of European pension funds, we of trends and developments that could influence our long-term expect a concomitant shift from unfunded duration hedging (using outlook in future years. derivatives) to funded duration hedging. The pace at which this occurs will be driven by the path of funding levels; as discussed earlier, we expect the shift, in general, to be slow and to vary greatly across individual pension funds. Some funds are already well down this path, but many have a long way to go.

38 LONG-TERM CAPITAL MARKET ASSUMPTIONS MATCHING CASH FLOWS AND MANAGING LIQUIDITY IN MATURING PENSION FUNDS

Pension fund demand for bonds may eventually fall away It is unlikely that public defined benefit and defined contribution plans will replace this demand for credit, as neither has the same The expected sustained demand for bonds from pension funds is imperative to de-risk and cash flow-match liability payments. The already well established. Indeed, matching assets and liabilities as precise shape of decumulation strategies for defined contribution an approach to de-risking pension plans is a long-standing and savings is a topic of much current debate as those systems move powerful industry megatrend. toward maturity, and will likely result in a mix of annuitization, However, we do begin to foresee some potentially tangible effects income-oriented asset strategies and managed divestment. Public arising should pension funds entering into runoff mode adopt buy defined pension plans will have little appetite to de-risk into lower and maintain strategies. A shift toward buy and maintain credit- risk, lower returning investment strategies and will likely continue oriented portfolios would, all else being equal, further reduce to lean hard on their long-term nature and the relatively younger liquidity in the marketplace as pension funds buy into these age of their participants to support return-oriented strategies. strategies. Once pension funds become fully funded and fully matched, demand for bonds will begin to drop off—the principal A new era of equitization? repayments will be applied to service cash flow and will not be The period since the financial crisis has been characterized available for reinvestment. A reasonable proportion of pension by de-equitization2 and expansion of the credit markets, fueled by funds could reach this status in the next 10 to 15 years. This is ultra-low interest rates and further supported by pension and especially true for corporations that make one-off final payments insurance demand for low-risk assets. It is perhaps beyond the to facilitate risk transfers or take other substantive de-risking horizon of our long-term assumptions, but as interest rates measures when the magnitude of payments required comes into a normalize, demand for bond assets from defined benefit plans reasonable range. We estimate the impact on capital markets falls off and the global savings glut reverses, perhaps we’ll enter a would be on a scale similar to that of central banks exiting new era of equitization. quantitative easing (QE) (Exhibit 5).

Could the eventual tapering of pension fund bond demand have as great an impact on capital markets as the unwinding of QE? EXHIBIT 5: CURRENT CENTRAL BANK BALANCE SHEETS AND DEFINED BENEFIT BOND ASSETS (USD, BILLIONS)*

U.S. UK Europe Japan

Central bank balance sheet 4,460 565 4,710 4,469

Total defined benefit pension assets 8,992 2,352 3,117 2,696

Defined benefit bond assets 3,327 1,277 1,221 884

Source: OECD Global Pension Statistics, Pension Protection Fund Purple Book 2016, Willis Towers Watson Global Pension Assets Study, J.P. Morgan Asset Management; data as of June 30, 2017. * Results for Europe are based on data for Finland, Germany, Italy, the Netherlands, Spain and Switzerland, with each country assumed to have 100% of pension assets in defined benefit plans, except for the Netherlands, which is reported by Willis Towers Watson as having 94% of pension assets in defined benefit plans.

2 De-equitization refers to the substitution of debt for equity, especially at the level of global capital markets. At the company level, it is the process whereby companies shift from equity financing to debt financing.

J.P. MORGAN ASSET MANAGEMENT 39 THE FUTURE PATH OF CHINESE INTEREST RATES

The cost of capital in China’s changing markets Hannah Anderson, Global Market Strategist, Global Market Insights Strategy Leon Goldfeld, CFA, Portfolio Manager, Multi-Asset Solutions

IN BRIEF • Identifying a benchmark interest rate, the risk-free cost of capital, and understanding how it might move in the future are essential for investors to price assets and assess likely future returns.

• We forecast a path for Chinese benchmark interest rates, exploring how ongoing financial liberalization in China—involving a shift from explicitly policy-driven interest rates toward a more market-driven system—will impact all interest rates, now and over our long-term forecast horizon.

• We use a natural rate of interest (r*) approach to model a theoretical, unsubsidized economic cost of capital for China, incorporating our economic forecasts and assumptions about the evolution of China’s financial system.

• The average of China’s most frequently referenced lending and deposit rates, in real terms, is currently about 300 basis points (bps) below where our natural rate estimate suggests it would be in a more liberalized environment.

• The interest rates available to investors in China have moved significantly closer to our natural rate estimate in recent years, and we forecast that over the next 10 to 15 years the gap will narrow further as market forces gain influence.

40 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE COST OF CAPITAL IN CHINA’S CHANGING MARKETS

SOLVING CHINA’S INTEREST RATE PUZZLE Interest rates in China have largely been driven by policy EXHIBIT 1: REAL GDP GROWTH AND AVERAGE REAL LENDING AND Investors often rely on a benchmark interest rate, the cost of DEPOSIT RATES (Y/Y % CHANGE, AVERAGE REAL LENDING AND † capital in its most basic form, to make investment decisions. DEPOSIT RATES ) Determining this rate in China can be confusing, both because Real GDP growth Average real lending and deposit rates† % (y/y) China maintains several interest rates—varying by the type of 16 financial activity—and because many of these rates have not reflected economic fundamentals. Along with knowing the benchmark rate, assessing its likely future behavior is critical, 12 since investor returns over time are composed of this rate plus a risk premium. China’s ongoing transition to a more market-driven 8 financial system obscures the future path of all interest rates and, most important, how a benchmark interest rate might move. To 4 aid investor understanding of interest rate behavior in China, present and future, we construct an estimate of an economically 0 determined cost of capital—the natural rate of interest (r*). We ’04 ’08 ’12 ’16 then present a framework for using r* to understand how the Source: CEIC, National Bureau of Statistics of China, J.P. Morgan Asset Managment; financial system liberalization process underway in China will data as of August 21, 2017. 1 impact future interest rates. † The average of the nominal interest rate on one-year RMB loans deflated by the investment price index (IPI), a measure calculated from the productive investment components of the fixed asset investment price index, and the nominal interest rate A FINANCIAL SYSTEM WITH CHINESE on one-year RMB deposits deflated by the non-food consumer price index (nfCPI). CHARACTERISTICS Liberalizing the financial system is a top priority for China’s China’s financial system developed in response to the country’s authorities in order to advance economic development, dampen economic goals. Policymakers determined that an investment-led financial speculation and support China’s integration into the economic growth model would make best use of China’s abundant global economy. An aspect of liberalization is giving market forces labor supply. The low corporate borrowing costs this model requires greater sway over the cost of capital; extending credit at a below- are financed by high levels of domestic savings. These low rates are economically efficient cost has spurred financial crises elsewhere, facilitated by government participation in the banking system, a something China aims to avoid. Understandably, authorities are regulatory environment aligned with policy goals and the implicit unwilling to let market forces potentially derail development 2 official backing of financial activities. Extending low-cost credit, goals, and they fear that rapid change could cause instability. mainly to large corporate borrowers, has proven economically However, allowing market forces to play a larger role will serve successful but has begun to fuel unproductive investments and additional goals, like discouraging wasteful investment, since a financial speculation. market-determined interest rate would likely be higher than In China, restrictions on financial activities and interest rates tend to current rates, and would ration capital more efficiently. depress returns to savers and enhance the benefits to borrowers. Published rates can demonstrate limited responsiveness to economic activity; nevertheless, market forces are gaining influence THE THEORY BEHIND r* (Exhibit 1). For example, the People’s Bank of China (PBOC) As noted, there is a disconnect among a theoretical interest rate liberalized, or stopped officially setting, the deposit rate ceiling in determined by China’s economic situation, the policy need for cheap 2015 and the lending rate floor in 2013, but these rates have still borrowing costs and currently available rates. This disconnect tended to follow published guidance since then. makes it difficult for investors to assess the true cost of capital and how it might evolve, complicating the consideration of Chinese assets. To aid investors in this situation, we construct an estimate of the natural rate of interest (r*) for China. 1 The authors are greatly indebted to the work of Dong He, Honglin Wang and Xiangrong Yu at the Hong Kong Monetary Authority, whose methodology we follow. 2 Low rates and high savings are related, as low returns on savings equal higher required savings levels, which anchor rates at low levels, as discussed in this edition in “How demographic change will affect savings, growth and interest rates,” J.P. Morgan Asset Management, 2018 Long-Term Capital Market Assumptions.

J.P. MORGAN ASSET MANAGEMENT 41 THE FUTURE PATH OF CHINESE INTEREST RATES

In economics, r* is the benchmark cost of capital, which over the growth, as we would have expected, and that this gap will likely long run equals nominal output growth (typically measured by narrow over our Long-Term Capital Market Assumptions (LTCMA) nominal GDP) at a level where inflation is stable and the economy forecast horizon. is operating at its potential—a crucial assumption for many economic models. This golden rule forms a useful guide to how Real r* has been persistently below real GDP growth, but is moving output and interest rates should move in relation to each other closer and is the basis of long-run forecasts for many assets. In emerging EXHIBIT 2: REAL GDP GROWTH AND REAL r* (Y/Y % CHANGE, ESTIMATED markets, r* is often well below the rate of nominal GDP growth, REAL r* ANNUAL %) partly because those financial systems are relatively inefficient at Real GDP growth Real r* % (y/y) intermediating credit and growth, but this gap narrows as financial 16 systems develop.3 Accessible interest rates for investors may be lower than r* due to policies that favor investment activity. 12 Therefore, we would expect first that China’s r* sits below nominal

GDP growth and, second, that the country’s rapid development in 8 recent decades would have narrowed this gap substantially.

Movements in the real natural rate of interest—r* adjusted for 4 inflation—significantly influence investment decisions.4 Since most 0 investors want to beat inflation, to at least preserve their money’s ’79 ’89 ’99 ’09 ’19 ’29 value, they care more about real interest rates. The benchmark Source: CEIC, National Bureau of Statistics of China, Penn World Tables, PwC, 5 interest rate, typically determined by a market’s key interest rate, J.P. Morgan Asset Management 2018 Long-Term Capital Market Assumptions; data as of guides investors when they consider how much compensation they August 21, 2017. require, at a minimum, after inflation, to invest instead of save. Real r* is estimated, here and throughout, utilizing the model described in this paper. Borrowing is relatively cheap, and higher returns may be found in investments vs. savings when the real key rate is below real r*. In Chinese economic growth could potentially be faster or slower theory, over the long run, the gap between real r* and the real than our base case. If so, the capital stock, and therefore real r*, key interest rate should equal zero in an economy in equilibrium. might also evolve differently. We remodeled our real r* estimate They are never perfectly equal in reality, but the gap should under reasonable lower and higher growth scenarios and average to zero over one business cycle. concluded that slower-than-expected growth would have little impact on the trajectory of real r*, while higher growth would place real r* below real GDP growth for the foreseeable future. GROWTH AND r* IN CHINA Authorities would likely have to turn on the credit spigots to achieve higher growth, as they did during the 2009 credit boom, Because r* is not an immediately observable variable, it must be resulting in a large unproductive debt load. The economic impact estimated from existing economic data. As economic conditions of that, and related policy decisions, would likely hold a shift over time, so does r*. Given that real r* is the more benchmark interest rate below real GDP growth. important rate for investors, we construct an estimate of real r* and its future path for China (see “Estimating China’s r*”). Comparing our estimate of China’s real r* with real GDP growth THE FUTURE OF FINANCIAL LIBERALIZATION (Exhibit 2) reveals that r* has been persistently below real GDP Chinese authorities recognize the need for more robust sources

3 Julio Escolano, Anna Shabunina and Jaejoon Woo, “The puzzle of persistently of financing—to efficiently support growth and ensure financial negative interest rate-growth differentials: Financial repression or income catch- stability—and have been reforming the system of allocating up?” International Monetary Fund Working Paper 11/260, 2011. https://papers.ssrn. com/sol3/papers.cfm?abstract_id=1961907 capital. Reforms will introduce more market forces, but the 4 John C. Williams, “FRBSF Economic Letter: The natural rate of interest,” Federal financial system will not become entirely independent of the Reserve Bank of San Francisco (2003). http://www.frbsf.org/economic-research/ government. Financial institutions have a strong preference for publications/economic-letter/2003/october/the-natural-rate-of-interest/ lending to government-linked projects, or those that appear to 5 The key interest rate in an economy determines the cost of money, which usually then determines the benchmark rates for borrowing and lending. In economies have government backing, due to both political concerns and the with interest-targeting central banks, like the U.S., the federal funds rate is the key low default risk. This, combined with state involvement in banking, interest rate. The PBOC is not an interest-targeting central bank and therefore does not set a benchmark rate for the economy as a whole. This is why we construct a has created an implicit guarantee for the entire financial system. benchmark rate of our own from an average of deposit and lending rates.

42 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE COST OF CAPITAL IN CHINA’S CHANGING MARKETS

Rightly or wrongly, investors perceive that most investment enjoys DRAWING INVESTMENT CONCLUSIONS FROM OUR a government backstop.6 Over the next 10 to 15 years, this implicit MODEL guarantee is likely to remain in effect. Removing it entirely would require a much higher tolerance for volatility than we believe This framework will be most useful to investors where it can be policymakers possess. applied to rates currently on offer. While we have used the average of lending and deposit rates as a logical point from which Consequently, we expect China’s financial system will continue to to estimate the size of the subsidy borrowers enjoy in China, in subsidize investment, leaving rates, even in a more market-driven practice these rates are not frequently used in trading. That leaves system, below our estimate of real r*. In practice, market- investors in need of a local reference rate. Limited data is determined rates should not equal real r*, since financial available on the history of China’s interest rates, and the institutions charge slightly above the benchmark interest rate to appropriate rate for investors to reference has shifted several cover their costs. We can measure present distortions between the times in recent history, a pattern we expect to continue. Currently, rates available and an estimated economically determined interest China’s seven-day repo rate is the most appropriate, accessible rate by comparing the average of real deposit and lending rates market rate to which we can apply our framework. with our estimate of real r*. Considering how the difference between these will evolve in the context of ongoing financial system Examining the relationship among the seven-day repo rate, reforms, we forecast a future deviation from real r*. adjusted for inflation, the average of real lending and deposit rates, and our estimate of real r* yields two important insights. In Exhibit 3, we demonstrate the current difference is around The seven-day repo rate and the lending and deposit rates’ 300bps. As market forces play a larger role in setting these rates, average have begun to converge, suggesting market forces are we anticipate that by the end of our 10- to 15-year window this gap indeed playing a larger role in setting interest rates. And the gap will have narrowed to between 50bps and 100bps. This forecast is between these two rates and real r* is narrowing (Exhibit 4). This based on our judgment of the pace and degree to which China’s indicates that investors can increasingly rely on real r* as a guide financial system becomes more market-driven. for the direction of future accessible interest rates, relative to what economic fundamentals suggest.

6 Douglas J. Elliot and Kai Yan, “The Chinese financial system: An introduction and overview.” (2013) The John L. Thornton China Center at Brookings. https://www. brookings.edu/wp-content/uploads/2016/06/chinese-financial-system-elliott- yan.pdf

We project that real benchmark interest rates will converge closer to an economic rate

EXHIBIT 3: REAL r* AND AVERAGE OF REAL LENDING AND DEPOSIT RATES (ANNUAL %, AVERAGE OF LENDING AND DEPOSIT RATES†)

Average of real lending and deposit rates† Real r* 10

0 ’04 ’10 ’16 ’22 ’28

Source: CEIC, National Bureau of Statistics of China, Penn World Tables, PwC, J.P. Morgan Asset Management 2018 Long-Term Capital Market Assumptions; data as of August 21, 2017. † The average of the nominal interest rate on one-year RMB loans deflated by the investment price index (IPI), a measure calculated from the productive investment components of the fixed asset investment price index, and the nominal interest rate on one-year RMB deposits deflated by the non-food consumer price index (nfCPI).

J.P. MORGAN ASSET MANAGEMENT 43 THE FUTURE PATH OF CHINESE INTEREST RATES

Market forces are having more influence on China’s interest rates EXHIBIT 4: REAL r*, AVERAGE OF REAL LENDING AND DEPOSIT RATES, AND REAL SEVEN-DAY REPO RATE (ANNUAL %, AVERAGE OF REAL LENDING AND DEPOSIT RATES,† REAL SEVEN-DAY REPO RATE††)

Average of real lending and deposit rates† Real seven-day repo rate†† r* % 8

0 ’02 ’05 ’08 ’11 ’14 ’17

Source: CEIC, National Bureau of Statistics of China, Penn World Tables, PwC, J.P. Morgan Asset Management; data as of August 21, 2017. † The average of the nominal interest rate on one-year RMB loans deflated by the investment price index (IPI), a measure calculated from the productive investment components of the fixed asset investment price index, and the nominal interest rate on one-year RMB deposits deflated by the non-food consumer price index (nfCPI). † † Seven-day repo rate is deflated by the average of the IPI and the nfCPI.

While we believe this paper outlines a useful tool for investors, a banks’ continued preference for supporting state-owned few complicating factors should be considered when drawing enterprises. But banks’ increasing willingness to finance those conclusions from the model. Our estimate is drawn from existing firms—typically smaller and privately held—engaged in the data sources, which are not entirely reliable. Our model assumes transition toward a services economy, plus the loosening of rate the economy is operating at its potential; we believe China could controls on different financial activities and the gradual removal of see productivity gains in excess of recent history. If this occurs, explicit policy guidance, will likely raise the average lending rate real r* would likely move toward real GDP growth at a faster pace above its current level. than we currently estimate. Nonetheless, the cost of investment is likely to remain below r* as a result of the implicit guarantee system. Financial liberalization RISKS REMAIN ON THE PATH OF FINANCIAL undertaken in these circumstances tends to decrease financial LIBERALIZATION stability, raising questions for policymakers about how to best ensure stability while meeting their reform goals.8 Even with these Higher interest rates—even if they are lower than fully market- questions outstanding, we expect investors’ attraction to China will determined rates—will further the development of China’s financial increase as its capital markets continue to develop. system. As rates rise, households and corporates will likely further diversify their portfolios beyond bank deposits, shrinking available The framework presented here offers investors a method for lending pools and forcing many banks to compete by offering higher evaluating the economic cost (for which real r* is a conservative deposit rates. Higher returns on savings, combined with the ongoing proxy) of the capital they might want to put to work in China, and cleanup of the financial system, will likely divert some assets away how that cost may rise over time. Comparing this model with the from currently popular but less well-regulated credit products. path of available rates in China can provide investors with a guide However, demand for higher returns—especially among the growing to an appropriate reference rate they should require for parting urban middle class—will likely continue to draw investors to riskier with their capital. These tools should enable more informed instruments. consideration of onshore assets in China’s changing financial landscape. Some of these assets will likely find their way into more traditional portfolio investments, supporting the development of China’s corporate debt and equity markets.7 Lending rates will likely remain lower than the natural rate would suggest, reflecting

7 This edition of our Long-Term Capital Market Assumptions features an expanded 8 Diego Anzoategui, Mali Chivakul and Wojciech Maliszewski, “Financial distortions in set of estimates for onshore Chinese assets. The expansion of China’s financial China: A general equilibrium approach,” International Monetary Fund Working Paper markets is also covered in our Fixed Income Assumptions. WP/15/274 (2015). https://www.imf.org/external/pubs/ft/wp/2015/wp15274.pdf

44 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE COST OF CAPITAL IN CHINA’S CHANGING MARKETS

ESTIMATING CHINA’S r * In any economy, the interest rate relates the cost of an investment of income, the depreciation rate for China’s capital stock calculated and its return. We can expand this principle into a model where a in the Penn World Table, our own estimate of China’s real capital country has a natural rate of interest that equates investment and stock, and annual real GDP in this equation to arrive at an estimate return. A country’s economic return, typically measured by GDP, of real r*. can be broken down into two components: labor and capital. A Estimating China’s capital stock, data that China does not publish, marginal-returns-to-capital function allows us to estimate what level was the most challenging part of this exercise. The essential of interest rates, r*, equates output and its components for China, question was: How much productive capital is created from the focusing on the relationship between investment—how capital is country’s high investment rate since not all investment goes built—and interest rates. We do this by determining the real return into economically productive activities? Different methodologies on the marginal product of capital—or how much an investor (here, produce vastly different results. We took two approaches: One anyone investing in China’s capital stock) will receive in proportion to assumed capital stock grows at the pace of real capital formation’s the additional economic output we can get from one additional unit contribution to real GDP, and the other assumed it grows at the rate of investment. of productive fixed asset investment (ex-financial and real estate This concept can also be written as: investment). The first likely overstated the present capital stock and the second likely underestimated it, so we used the average of the r* = (α/(K/Y) – δ) * (1- τ) produced real r* estimates for further analysis. Looking forward, Where α is capital’s share of output, K is the real capital stock, δ is we assumed that the recent average discrepancy between the rate the depreciation rate on that capital stock, Y is real output, and τ is of real GDP and capital stock growth held in each methodology and the tax rate on capital.i We use China’s corporate income tax rate incorporated our macroeconomic assumption for real GDP growth.ii for taxes on capital, gross fixed capital formation’s share of GDP as capital’s share i Dong He, Honglin Wang and Xiangrong Yu, “Interest rate determination in ii The starting capital stock was estimated using data and methodologies China: Past, present, and future,” International Journal of Central Banking presented by Gregory C. Chow, “Capital formation and economic growth (2015). http://www.ijcb.org/journal/ijcb15q5a7.htm in China,” Quarterly Journal of Economics 108:3 (1993). https://www. jstor.org/stable/2118409?seq=1#page_scan_tab_contents, and by Kui- Wai Li, “China’s Capital and Productivity Measurement Using Financial Resources,” Yale University Economic Growth Center Discussion Paper No. 851. https://ssrn.com/abstract=382400

J.P. MORGAN ASSET MANAGEMENT 45 U.S. DOLLAR FORECAST

The path of the U.S. dollar: Looking forward by looking back Dr. David Kelly, CFA, Chief Global Strategist, Head of Global Market Insights Strategy Roger Hallam, Chief Investment Officer,Currency Management John C. Manley, Global Market Strategist, Global Market Insights Strategy

IN BRIEF The U.S. dollar (USD) exchange rate is a critical variable shaping long-term asset returns. Presently, the currency is overvalued in real terms, which is likely unsustainable in the long run. We expect the dollar to be pushed gradually lower over our Long-Term Capital Market Assumptions forecast period by:

• a large U.S. current account deficit

• narrowing interest rate and real economic growth differentials between the U.S. and its major trading partners, resulting in relative capital flows supportive of non-dollar currencies

• the U.S.’s gradual abandonment of its several-decades old “strong dollar policy”

46 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE PATH OF THE U.S. DOLLAR: LOOKING FORWARD BY LOOKING BACK

INTRODUCTION A visual history of the modern U.S. dollar EXHIBIT 1: NOMINAL MAJOR CURRENCY TRADE-WEIGHTED EXCHANGE RATE, One critical variable shaping long-term global financial asset returns U.S. DOLLAR, MONTHLY, INDEXED is the exchange rate of the U.S. dollar. This is, of course, obvious for 150 U.S. residents investing abroad and for foreign investors in U.S. 140 assets, but beyond the currency-hedging implications, it is also an 130 important driver of U.S. inflation and interest rates and has a 120 powerful impact on global commodity prices. 110

But what fundamentally drives the U.S. dollar? Three broad forces: 100

Classical economic theory highlights the importance of balance of 90 payment flows and suggests that, in the long run, purchasing power 80 87.1 parity and current account balances should drive the currency. A 70 more monetarist view focuses on financial flows: A country that attracts foreign capital through higher interest rates or higher 60 ’73 ’84 ’95 ’06 ’17 expected returns on investments should, in theory, generally see an appreciation of its currency over time. In addition, bouts of active Source: Federal Reserve, FactSet, J.P. Morgan Asset Management; data as of September 30, 2017. policy intervention to manipulate exchange rates have been overlaid on this markets-based framework by both governments and central banks, either individually or in coordination. Trade and the dollar: The aftermath of Bretton Woods

In our Long-Term Capital Market Assumptions, we take a long look Under the Bretton Woods agreement, signed in 1944, a global forward at how key economic variables and financial markets may monetary management system was installed for the first time in evolve over the next 10 to 15 years. When attempting this for the history: Signatory countries pledged to peg their respective U.S. dollar, we start by looking at how these three broad forces currencies to gold, while the U.S. government guaranteed a fixed have shaped the path of the dollar in recent decades, then we rate conversion of gold to U.S. dollars—$35 per ounce—effectively look at how these forces may evolve in the future and why we maintaining an exchange rate across the developed world.2 This believe they justify projecting a generally declining dollar over our was not a pure gold standard, which would have constrained global forecast period. money supply growth to the very slow growth in the gold stock. Still, the guarantee of convertibility at the heart of the Bretton Examining 45 years of twists and turns in the dollar allows us to Woods framework became increasingly untenable in the 1960s, as explain chronologically the broad forces that have shaped its path. nominal GDP growth, and consequently the money supply, far Looking at the nominal major currency index for the U.S. dollar outstripped the growth in the global gold stock. from January 1973 to September 2017 (Exhibit 1), we see how extreme and long-lasting currency regimes can be.1 After the U.S. and other countries abandoned direct convertibility in the early 1970s, the U.S. dollar became a floating currency. At first, in the otherwise tumultuous early 1970s, the U.S. dollar exchange rate was relatively stable against major U.S. trading partners. Later in the decade, however, the U.S. dollar began to drift downward. This move largely reflected the traditional and predictable impact of balance of payment flows: In a decade of very high inflation throughout the developed world, U.S. inflation managed to be worse, on net, with the effect of gradually shifting the U.S. from maintaining a current account surplus to a current account deficit (Exhibit 2). This current account deficit, in turn, increased the international supply of dollars, putting downward 1 The Federal Reserve’s major currency index is a geometrically weighted average pressure on the exchange rate as dollar supply ballooned. of the U.S. dollar’s bilateral exchange rates with seven currencies: the euro (and its predecessors), the Canadian dollar, the Japanese yen, the British pound, the Swiss franc, the Australian dollar and the Swedish krona. The weights are based on exports, imports and third-country trade for each currency region. See “Indexes of 2 Signatory countries included the U.S., Canada, Australia, Japan and Western Europe the foreign exchange value of the dollar,” Federal Reserve Bulletin (Winter 2005), (Germany, France, the UK, Italy, Spain, the Netherlands, Belgium, Switzerland, https://www.federalreserve.gov/pubs/bulletin/2005/winter05_index.pdf Greece, Denmark, Finland and Norway).

J.P. MORGAN ASSET MANAGEMENT 47 U.S. DOLLAR FORECAST

The U.S. maintains a substantial current account deficit Policy intervention and the dollar: The 1985 Plaza Accord EXHIBIT 2: U.S. CURRENT ACCOUNT BALANCE AS A % OF NOMINAL GDP, QUARTERLY SAAR The Reagan administration initially regarded the rising dollar of % the early 1980s as a benign reflection of greater confidence in the 2 American economy. This sentiment was reversed, however, after President Ronald Reagan’s re-election in 1984. James Baker, the 0 new Treasury secretary, signaled that the Treasury would need to re-examine its previous policy of non-intervention in exchange -2 rates to address growing U.S. concern about a rising dollar. This -2.6 re-examination led to the 1985 Plaza Accord, in which, at New -4 York’s Plaza Hotel, the G5 nations (the U.S., Japan, West Germany, France and the UK) agreed to intervene jointly to push the dollar -6 down. While the central banks’ actual currency transactions were relatively modest, the joint nature of their actions sent markets a -8 ’67 ’77 ’87 ’97 ’07 ’17 powerful signal. The dollar had already declined from its peak, reached six months before the accord was signed as the coming Source: U.S. Bureau of Economic Analysis, FactSet, J.P. Morgan Asset Management; data as of September 30, 2017. policy changes were being signaled. It fell more sharply following the September 22 statement, retreating 39% by April 1988.

Financial flows and the U.S. dollar: The early 1980s Over the years, many economists have doubted the impact of central bank interventions on foreign currencies, particularly if the The U.S. recessions of the early 1980s were “stagflationary,” actions are “sterilized”—that is, if foreign currency transactions are characterized by simultaneously high inflation and high offset domestically to prevent any impact on the domestic money unemployment. To combat the inflation component, which had supply. However, the Plaza Accord experience suggests the been unacceptably high for over a decade, the Federal Reserve opposite: When a market is dangerously caught up in rampant (Fed) under then-chairman Paul Volcker implemented a very tight speculation and momentum, central bank messages can be very monetary policy, boosting nominal U.S. interest rates to extreme effective, provided they are trying to push the currency in an highs. In addition, tax cuts and increased defense spending economically logical direction and the policy’s purpose is clearly introduced by the Reagan administration early in the decade signaled to markets. added to the federal deficit but also helped pull the U.S. out of recession, eventually leading to outsized economic growth. Financial flows and a “strong dollar policy”: Extremely high U.S. interest rates attracted global investment into The 1990s U.S. dollar revival U.S. bonds, while the promise of stronger economic growth fueled By early 1987, a severely depreciated dollar following the Plaza international flows into U.S. equities and real estate. The net result Accord had not resulted in a contraction of the U.S. current was that, although the U.S. still had a significant current account account deficit, nor had Japan or Germany seen their current deficit in the early 1980s, the dollar began to appreciate account surpluses decline. Economists opined that a falling dollar drastically, rising 55% from July 1980 to March 1985. hadn’t fixed the U.S. current account deficit due to a large U.S. In retrospect, this appreciation seems wildly overdone. The U.S. budget deficit—America was consuming more than it was was far from the only attractive opportunity for global investors in producing because the government was spending in excess of its the 1980s, and because of the lagged effect of the exchange rate revenues. This was frequently referred to as the “twin deficit” on current account economics, the U.S. dollar peaking in 1985 problem (Exhibit 3). would very likely have led to a much bigger trade deficit in the With the benefit of hindsight, we can see that critics should perhaps years that followed, pushing the currency back down. However, have been a little more patient—exchange rates impact the current investors had become irrationally convinced that the U.S. dollar account balance with a lag, and the U.S. current account deficit did was a one-way bet, contributing to a speculative spike in the improve in the late 1980s, even as the fiscal deficit worsened. exchange rate. Nevertheless, fiscal mismanagement has always been a popular scapegoat, particularly for global finance ministers and financial institutions. Consequently, in February 1987 with the Louvre Accord, six nations (the G5 signatories of the Plaza Accord plus Canada)

48 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE PATH OF THE U.S. DOLLAR: LOOKING FORWARD BY LOOKING BACK

The U.S. has faced a “twin deficit” problem EXHIBIT 3: U.S. CURRENT ACCOUNT BALANCE AND U.S. FISCAL BALANCE AS A % OF NOMINAL GDP, ANNUAL Fiscal balance Current account balance % 4

-2

-4

-6

-8

-10 ’68 ’71 ’74 ’77 ’80 ’83 ’86 ’89 ’92 ’95 ’98 ’01 ’04 ’07 ’10 ’13 ’16

Source: U.S. Bureau of Economic Analysis, Congressional Budget Office, FactSet, J.P. Morgan Asset Management; data as of September 30, 2017. agreed to a trading range for the U.S. dollar3 to try to stop its thus- alongside booming economic growth and, eventually, higher U.S. far relentless slide. The accord also prescribed a reduced federal interest rates, combined to push the dollar up 40% between April deficit for the U.S. and urged Japan and Germany to take measures 1995 and February 2002. This, in turn, contributed to a rising to reduce their current account surpluses. current account deficit, which reached almost 6% of GDP in 2006.

There was little reason to believe that the Louvre Accord would be This high dollar, which had clearly broken away from the anchors fruitful; indeed, the agreement did nothing to prevent the dollar of economic fundamentals, finally began to retreat in the mid- from falling a further 12% over the next 14 months. More 2000s, but not before dealing a severe blow to the international significantly, the October 1987 stock market crash was widely competitiveness of most of the U.S. manufacturing sector, blamed on distortions caused by the earlier Plaza Accord, casting a problem that still resonates today. coordinated central bank intervention in currency markets as a chaotic and negative force. This change of sentiment toward By March 2008, the dollar had fallen back to levels that might, in intervention is critical in explaining the dollar’s trajectory since time, have brought the current account deficit close to the late 1980s: Without a popular desire for intervention, the U.S. equilibrium. However, the unfolding subprime mortgage crisis would tolerate a highly over-valued currency for long periods over increased fears in global markets and—somewhat perversely, since the next 30 years. the crisis was centered in the U.S.—caused a global flight to safety, boosting demand for U.S. Treasuries and, in turn, for the U.S. From the failure of the Louvre Accord until the mid-1990s, the U.S. dollar. This “fear trade” reversed soon after the stock market dollar remained relatively range-bound, and by 1991 the U.S. had troughed in March 2009, but re-emerged in 2011 with the managed to post a small current account surplus for the first time eurozone crisis, again nudging the dollar higher. in a decade. This would be short-lived. In the late 1990s, the dollar began to rise more quickly. The economy started to boom as The dollar was bid up further in 2014 and 2015, prompted by the technology investment spending surged and the Clinton perception that the U.S. was once again outpacing its trading administration, determined to avoid being labeled too “liberal” on partners in growth and by the expectation that the Fed would financial issues and willing to use a rising current account deficit soon boost interest rates. Finally, in 2016 and up to the present, as a safety valve to defuse concerns about an overheating the importance of expectations was again underlined as a economy, maintained the policy position that “a strong dollar is in strengthening global economy, and a lack of confidence in Fed the interest of the United States.”4 This strong dollar policy, resolve, pushed the dollar sideways and then down—even as the Fed began to implement the tightening that it had long threatened. A brief spike in the dollar following the U.S. 3 Jeffrey Frankel, “The Plaza Accord, 30 years later,” NBER Working Paper 21813 (December 2015). http://www.nber.org/papers/w21813 presidential election, predicated on rhetoric heralding U.S. 4 David E. Sanger, “Rubin hints at end to long U.S. push of strong dollar,” New York economic growth, dissolved as the promised policy change failed Times (February 1997). http://www.nytimes.com/1997/02/08/business/rubin- hints-at-end-to-long-us-push-of-strong-dollar.html

J.P. MORGAN ASSET MANAGEMENT 49 U.S. DOLLAR FORECAST wwwwwto materialize and efforts got underway to reduce U.S. Future U.S. inflation should exceed that of most major trading trade imbalances, resulting in a 2017 that has thus far brought the partners dollar closer to fair value. EXHIBIT 5: U.S. HEADLINE CPI Y/Y CHANGE VS. TRADE-WEIGHTED HEADLINE CPI Y/Y CHANGE OF MAJOR TRADING PARTNERS The dollar going forward 2.3% So where will these forces likely drive the dollar? At present, the 1.7% dollar is modestly ahead of historical long-term real valuations, 1.5% and about 10% above our ex ante estimate of fair value (Exhibit 1.3% 4). It has moved higher as a result of the global economic crisis and tolerant U.S. policy. This overvaluation is likely unsustainable in the long run.

The U.S. dollar is currently modestly overvalued relative U.S. MAJOR TRADING PARTNERS U.S. MAJOR TRADING PARTNERS to history 10-year average Forecast (10- to 15-year average)

EXHIBIT 4: REAL MAJOR CURRENCY TRADE-WEIGHTED EXCHANGE RATE, Source: U.S. Bureau of Labor Statistics, Statistics Canada, Melbourne Institute, Swiss U.S. DOLLAR, MONTHLY, INDEXED Federal Statistical Office, SCB—Statistics Sweden, Japan Ministry of Affairs & 120 Communications, UK Office for National Statistics, Eurostat, FactSet, U.S. Bureau of Industry and Security, J.P. Morgan Investment Bank, J.P. Morgan Asset Management; 10-year average major trading partners data include J.P. Morgan estimates for Canada 110 and Japan September headline inflation; data as of September 30, 2017. Forecasts based on 2018 Long-Term Capital Market Assumptions data.

100 99.7 Second, financial flows should also generally push the dollar down. As of September 2017, both long-term and short-term interest rates 90 Avg.: 92.3 are higher in the U.S. than in most of its major trading partners. To some extent, this simply reflects that the U.S. is further along in its 80 economic expansion than its trading partners, although we expect U.S. rates will continue to be relatively higher in the decade ahead 70 ’92 ’98 ’04 ’10 ’16 (Exhibits 6 and 7). That said, we are likely beyond the peak of central bank policy divergence and, as such, short-term yield Source: Federal Reserve, FactSet, J.P. Morgan Asset Management; data as of September 30, 2017. differentials are likely to narrow, providing less support for the dollar compared with the present. Our forecasts show a modest widening of the gap in nominal long-term bond yields, but because First, balance of payment flows should generally push the dollar of faster rising U.S. inflation, the gap in real long-term bond yields down. In the second quarter of 2017, the U.S. current account should narrow, putting further downward pressure on the dollar. deficit was 2.6% of GDP, or roughly $500 billion annualized. This is gradually pumping dollars into the world economy and should, Additionally, the real economic growth differential between the by creating an oversupply, push the dollar lower. U.S. and its developed market trading partners should narrow over the next 10 to 15 years (Exhibit 8) and, given more Inflation differentials should also reduce the nominal value of the compelling relative valuations, we expect international equities to dollar. In our Long-Term Capital Market Assumptions, we assume provide generally better returns than U.S. equities over that time that U.S. inflation will exceed that of most of its major trading horizon. The resulting relative capital flows should provide support partners5 by about 0.7% per year (Exhibit 5). for non-dollar currencies. An improvement in European institutional infrastructure in recent years should also have a similar impact on the euro-dollar relative value; in the years ahead, we will likely see at least a partial reversal of the outflows 5 Major trading partners are defined as the countries/regions included in the Federal Reserve’s major currency index: Canada, Australia, Switzerland, Sweden, Japan, from Europe that occurred during the eurozone crisis (Exhibit 9). the UK and the eurozone. Aggregate figures are calculated taking the given metric (i.e., inflation) and weighting it by trade weights as provided by the U.S. Bureau of Industry and Security. Current figures are as of September 2017. This methodology is used throughout this paper whenever “trade-weighted major trading partners” is referenced.

50 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE PATH OF THE U.S. DOLLAR: LOOKING FORWARD BY LOOKING BACK

The short-term interest rate gap between the U.S. and its major The 10-year interest rate gap between the U.S. and its major trading partners should narrow moving forward trading partners should widen modestly

EXHIBIT 6: U.S. 3-MONTH BENCHMARK BILL YIELD VS. U.S. 3-MONTH EXHIBIT 7: U.S. 10-YEAR BENCHMARK BOND YIELD VS. 10-YEAR YIELD BENCHMARK BILL YIELD OF MAJOR TRADING PARTNERS OF MAJOR TRADING PARTNERS 3.5%

2.7% 2.3% 2.3% 1.6%

1.1% 1.8%

0.2% U.S. MAJOR TRADING PARTNERS U.S. MAJOR TRADING PARTNERS U.S. MAJOR TRADING PARTNERS U.S. MAJOR TRADING PARTNERS Current Forecast (10- to 15-year average) Current Forecast (10- to 15-year average)

Source: Tullett Prebon, Statistics Canada, Australian Financial Markets Association, Bank Source: Tullett Prebon, FactSet, U.S. Bureau of Industry and Security, JPMorgan of Sweden, ICE Benchmark Administration, FactSet, U.S. Bureau of Industry and Security, Chase & Co., J.P. Morgan Asset Management; data as of September 30, 2017. J.P. Morgan Asset Management; data as of September 30, 2017. Forecasts based on 2018 Forecasts based on 2018 Long-Term Capital Market Assumptions data. Long-Term Capital Market Assumptions data.

Third, policy intervention may push the dollar down in the years is now more than 30 years old, history suggests that a determined ahead. For the last 30 years, under both Republican and and visible attempt by the U.S. to depreciate an overvalued dollar Democratic administrations, the U.S. has generally pursued a can work. so-called “strong dollar policy,” despite its negative impact on U.S. From the dollar’s presently overvalued state, it is reasonable to manufacturing. However, the presidential election of 2016 may assume that the currency should follow a path laid out by mark a turning point in U.S. trade policy: The Trump fundamental forces. Together, these forces—including the U.S.’s administration is more explicitly protectionist than any of its significant current account imbalance, a convergence in global recent predecessors. Moreover, while this protectionism has often interest rates and economic growth rates and the gradual been manifested specifically in calls for tariffs or “border abandonment of the U.S.’s strong dollar policy—underlie our adjustments,” a more general stance of encouraging a weaker assumption of a gradual decline in the U.S. dollar exchange rate dollar may yet prove more palatable at home and harder to over the next 10 to 15 years. oppose around the world. While the best example of such a policy

The economic growth gap between the U.S. and its major trading The euro has lost status against the dollar as a reserve partners should narrow currency, but this should change EXHIBIT 8: REAL U.S. GDP GROWTH VS. TRADE-WEIGHTED REAL GDP EXHIBIT 9: % OF GLOBAL RESERVES ALLOCATED TO THE EURO AND U.S. GROWTH OF MAJOR TRADING PARTNERS, SAAR DOLLAR

2.2% % allocated to USD (LHS) % allocated to EUR (RHS) 74 29 1.8% 72 27

1.4% 1.3% 70 25

68 23

66 21

64 19

62 17 U.S. MAJOR TRADING PARTNERS U.S. MAJOR TRADING PARTNERS 60 15 5-year average Forecast (10- to 15-year average) ’99 ’05 ’11 ’17

Source: U.S. Bureau of Economic Analysis, Statistics Canada, Australian Bureau of Statistics, Source: International Monetary Fund, FactSet, J.P. Morgan Asset Management; data Swiss State Secretariat for Economic Affairs, SCB—Statistics Sweden, Japanese Cabinet as of September 30, 2017. Office, UK Office for National Statistics, Eurostat, FactSet, U.S. Bureau of Industry and Security, J.P. Morgan Asset Management; data as of September 30, 2017. Forecasts based on 2018 Long-Term Capital Market Assumptions data.

J.P. MORGAN ASSET MANAGEMENT 51

II Assumptions FIXED INCOME ASSUMPTIONS

G4 government bonds: Slow road, low yields John Bilton, CFA, Head of Global Multi-Asset Strategy, Multi-Asset Solutions Michael Feser, CFA, Portfolio Manager, Multi-Asset Solutions Jonathon Griggs, Head of Applied Research, Global Fixed Income, Currency & Commodities Sorca Kelly-Scholte, FIA, Global Strategist, Global Pension Solutions Nandini Srivastava, Quantitative Analyst, Multi-Asset Solutions Timothy Lintern, Analyst, Global Strategist, Multi-Asset Solutions

IN BRIEF • Expectations of an extremely slow path of normalization and a lower terminal rate are playing out even as global central banks become incrementally more hawkish.

• We make few changes to our economic framework this year, and continue to expect G4 10-year yields to reach equilibrium at or just below national nominal GDP.

• We expect the effects of regulation and aging populations to increasingly manifest themselves in ultra-long dated yields, especially for countries like the UK, where pension scheme underfunding challenges macro fundamentals as the principal driver of long- dated yields.

• Credit remains a bright spot in fixed income, and although the current credit cycle is rather mature, our long-term projections of credit spreads, defaults and recovery rates continue to imply a reasonable return uplift above government bonds.

• The emerging market (EM) debt outlook continues to improve, implying that it offers an attractive diversifier to credit portfolios even with current spreads quite close to long-term fair value.

54 LONG-TERM CAPITAL MARKET ASSUMPTIONS G4 GOVERNMENT BONDS: SLOW ROAD, LOW YIELDS

INTRODUCTION We expect a slow and shallow path of rate normalization across developed economies In last year’s edition of our Long-Term Capital Market Assumptions EXHIBIT 2: FORECAST FOR USD, EUR, GBP AND JPY CASH RATES

(LTCMAs), we significantly extended the time horizon over which we USD EUR GBP JPY anticipate normalization of interest rates to play out. We noted that % zero and negative nominal interest rate policies are unlikely to 3.0 remain a permanent feature over our forecast horizon. But with potential growth lower and little sign of inflation, there are few 2.0 reasons for policymakers to rush normalization. Over the last 12 months, we have seen increasing evidence of this “gradualism” 1.0 among major central banks. While the eight U.S. rate hiking cycles between 1970 and 2006 lasted, on average, 22 months, the current 0.0 cycle is already almost two years old, yet perhaps only halfway complete (Exhibit 1). Global central bank communications have -1.0 certainly taken on a more hawkish tone over the past 12 months, ’17 ’20 ’23 ’26 ’29 ’32 but the U.S. has arguably advanced the furthest down the path of hiking. Eventually, we expect rate normalization to occur in all Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. economies, but the path of normalization will be both slow and * Cash rates at equilibrium are not an estimate of the terminal rate for the current cycle, which may exceed the long-run equilibrium fair cash rate. The equilibrium shallow (Exhibit 2). cash rate represents the central tendency of policy rates over the long run.

Past U.S. rate hiking cycles lasted around 22 months, on average EXHIBIT 1: U.S. RATE HIKING CYCLES, 1972-PRESENT

Starting federal Ending federal Number of hikes Hiking cycle funds rate (%) funds rate (%) (in 25 bps increments) Number of months 1972-1974 3.50 13.00 38 28 1976-1979 4.75 15.50 43 36 1980-1981 10.00 20.00 40 10 1983-1984 8.50 11.75 13 16 1986-1989 6.00 9.63 15 31 1994-1995 3.25 6.00 11 13 1999-2000 5.00 6.50 6 12 2004-2006 1.25 5.25 16 25 Average 5.28 10.95 23 21 2015-Present 0.125 1.125* 4* 22 Source: Bloomberg; data as of September 30, 2017. * As of September 30, 2017.

OUR FIXED INCOME ASSUMPTIONS METHODOLOGY CONSTRUCTS EQUILIBRIUM YIELDS FROM SIMPLE BUILDING BLOCKS BUILDING BLOCKS: ANATOMY OF FIXED INCOME YIELDS AND SPREADS 1. Equilibrium cash rate 3. + Credit spread • The level of cash rates consistent with our long-run growth • Additional credit spread, incorporating rating migration and inflation forecasts by country assumptions for investment grade (IG) and credit/liquidity risk 2. + Curve (equilibrium long-dated yield) premia and expected default loss for high yield (HY) • Additional yield to compensate investor for holding long-term 4. Return calculation bonds (term premium) • Reflects normalization path to equilibrium interest rate, annual roll-down and rebalancing to a constant maturity index, plus coupon accrual and any defaults/losses

J.P. MORGAN ASSET MANAGEMENT 55 FIXED INCOME ASSUMPTIONS

We assume that a rank ordering of short rates will reflect the rates a little above equilibrium CPI despite a weakening growth rank ordering of real GDP growth across developed economies outlook—accounting for a modest political and inflation risk EXHIBIT 3: 10-YEAR AND 3-MONTH NOMINAL YIELDS VS. IMPLIED premium, given the uncertainty hanging over the UK economy for NOMINAL GDP, LTCMA 2018 ASSUMPTIONS the early part of our horizon. Aside from these outliers, the rank 10-yr 3-mo ordering of our baseline assumptions of real r*, which are near- 5.00 zero across developed economies, matches the rank ordering of our real GDP growth assumptions (Exhibit 3). 4.00 The elevated correlation of 10-year G4 bonds suggests that the 3.00 R² = 0.93178 globalization of longer-end yields persists as a market factor. The normalization of bond yields has begun, however, and this year’s 2.00 higher starting yields result in a pickup in duration premium Nominal yields % 1.00 R² = 0.91803 relative to cash. Even then, the duration premium remains quite compressed. Until G4 central banks have made further progress 0.00 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50 toward balance sheet normalization—and the term risk premium Implied nominal GDP % has picked up markedly (Exhibit 4)—it is hard to see projected bond returns rising meaningfully above cash returns. This also Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. implies flatter curves in equilibrium2 as the combined effect of For each region, implied nominal GDP is calculated by adding our forecast for core inflation to our forecast for real GDP. In the U.S., the inflation metric is PCE; for all aging populations, increased regulatory requirements to hold other regions, the inflation metric is CPI. bonds and larger central bank balance sheets weighs on longer- dated rates. Cash vs. 10-year curves will likely be flatter than historical averages, and the tendency should be especially acute in In recent editions of the LTCMAs, we have acknowledged the the very long end of yield curves, leading to quite flat 10-year vs. gradual decline in the estimated real natural short-term rate of 30-year curves across developed economies. interest (real r*) across major economies. For most major economies, we assume that real r* is a little above zero,1 in turn The persistently disappointing return outlook for sovereign bonds implying that equilibrium short-term interest rates are equal to or certainly does not eradicate all opportunities for bond investors. slightly above our equilibrium estimate of consumer price index Credit remains a bright spot, and despite growing cyclical (CPI) inflation. This year we maintain this assumption but also pay concerns we believe current spreads, despite being tighter than more attention to the relative ranking of real GDP relative to equilibrium fair value, provide for a reasonable return uplift given equilibrium interest rates across developed economies. We our long-term expectations of default and recovery rates. As a explicitly reduce Japanese equilibrium short rates below result, long term, default-adjusted returns across both high yield equilibrium inflation—implying a negative real rate of interest credit and EM debt look reasonably attractive. throughout our forecast horizon. By contrast, we see UK short

1 Thomas Laubach and John C. Williams. 2015. “Measuring the Natural Rate of Interest Redux.” Federal Reserve Bank of San Francisco Working Paper 2015-16. 2 Although in reality yield curves will periodically flatten and steepen in response http://www.frbsf.org/economic-research/publications/working-papers/wp2015-16. to shorter-term economic cycles, we have assumed a smooth evolution during the pdf adjustment process toward our assessment of long-term equilibrium 10- and 30- year yields and short-term policy rates.

Until the term risk premium has picked up substantially, it is difficult to see projected bond returns rising much above cash returns EXHIBIT 4: 10-YEAR TREASURY YIELD AND SUBSEQUENT 15-YEAR RETURN DIFFERENTIAL VS. CASH, HISTORICAL AND PROJECTED (%)

Starting 10-yr yield Cash 10-yr return di erential % 16

0 ’80 ’86 ’92 ’98 ’04 ’10 ’16

Source: J.P. Morgan Asset Management; estimates as of September 30, 2017.

56 LONG-TERM CAPITAL MARKET ASSUMPTIONS G4 GOVERNMENT BONDS: SLOW ROAD, LOW YIELDS

U.S. RATES In contrast to cash rates, U.S. 10-year yields are anchored a little closer to international paths of normalization* U.S. rates are following a normalization path roughly in line with a EXHIBIT 5: G4 10-YEAR YIELD NORMALIZATION hiking cycle lasting around four years. Given this trajectory, we USD EUR GBP JPY simply roll our baseline normalization assumptions forward by one Yields year and maintain our equilibrium cash rate assumption of 4.0 2.25%—the same level as CPI inflation and roughly 25 basis points 3.0 (bps) above personal consumption expenditures (PCE) inflation.

Of course, there is every likelihood that the terminal rate in this 2.0 hiking cycle will be above our equilibrium estimate. Nevertheless, 1.0 once normalization has occurred, we expect U.S. cash rates over the remainder of our forecast horizon to average 2.25%. Even 0.0 though cash rates will fluctuate around this average with the ’17 ’20 ’23 ’26 ’29 ’32 cycle, we maintain a U.S. 10-year yield estimate of 3.50%, which is 50bps below our forecast of nominal GDP growth and in line with Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. patterns over the last 15 years.

The globalization of bond yields anchors U.S. 10-year yields a little EUROZONE RATES closer to international paths of normalization, in contrast to cash rates, which are more domestically driven (Exhibits 5 and 6). We modestly upgrade our expectations of eurozone real GDP However, because normalization has started in the U.S. in the last growth, but we do not anticipate normalization of monetary policy 12 months, the higher starting point of bond yields results in a to begin until 2019. We expect that inflation will, on average, modest reversal of the recent spread compression between long- undershoot targets over our forecast horizon and thus hold our run estimates of cash and 10-year bond returns. equilibrium yield3 estimates steady, with cash at 1.75%, the 10-year at 3.00% and the 30-year at 3.25%. In common with U.S. patterns, we’ve held trajectories for both cash and bond yield normalization static, implying we are one year closer to normalization; even then, we do not expect EUR yields to reach equilibrium until 2023.

3 We use eurozone yields based on the French government curve as a benchmark.

Lower equilibrium yield and return assumptions reflect expectations of very gradual rate normalization, leading to a lower terminal rate EXHIBIT 6: DEVELOPED MARKET EQUILIBRIUM YIELD AND RETURN ESTIMATES (10-15 YEAR RETURN ASSUMPTIONS, LOCAL CURRENCY, %) USD GBP EUR JPY 2018 Equilibrium Return Equilibrium Return Equilibrium Return Equilibrium Return Yield Yield Yield Yield Inflation 2.25 - 2.00 - 1.50 - 1.00 - Cash 2.25 2.00 2.25 1.75 1.75 1.25 0.50 0.25 10-year bond 3.50 2.75 3.00 1.75 3.00 1.50 1.25 0.50 30-year bond 3.75 2.50 3.00 1.00 3.25 1.00 1.75 0.25 Gov’t bond market* 3.25 3.00 3.00 1.00 3.00 1.50 1.25 0.50 Investment grade credit** 5.00 3.50 4.50 2.50 3.75 2.00 2.00 1.00 High yield 8.25 5.25 6.75 3.50 Emerging market debt*** 6.75 5.25

Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. * U.S Intermediate Treasuries, UK Gilts, euro government bonds, Japanese government bonds. ** Investment grade corporate bonds. *** Emerging market sovereign debt.

J.P. MORGAN ASSET MANAGEMENT 57 FIXED INCOME ASSUMPTIONS

UK RATES GLOBAL CREDIT MARKETS: A BRIGHTER SPOT IN

The uncertainty created by the Brexit negotiation introduces a EXTENDED FIXED INCOME number of complicating factors for UK rates this year. Our real Despite the maturity of the current credit cycle, for a long-term growth estimates already reflect a moderate penalty to investor credit remains one of the more promising areas in the fixed equilibrium trend growth relative to the eurozone. It has long income universe. Our return estimates naturally reflect the currently been common for UK policymakers to rely on pound sterling to act compressed level of credit spreads, but we conclude that credit as a shock absorber during times of macro stress, and the offers an attractive return pickup relative to government bonds, aftermath of the UK’s vote to leave the EU is no different. This with U.S. investment grade and high yield offering returns of 3.50% brings some upside risk to inflation, and while we expect UK CPI to and 5.25%, respectively. find equilibrium close to the Bank of England’s 2.00% target, we believe some modest inflation and political risk premium need to The higher level of starting risk-free rates is a consideration be reflected in UK short-end rates. As a result, we expect the UK across credit markets. IG credit has a much greater duration than cash rate to be rather higher than equilibrium real GDP might HY credit and is therefore more sensitive to this move in the suggest, and for UK cash rates to normalize a little faster than underlying risk-free rates. For both IG and HY credit, the higher their eurozone neighbors. starting level of risk-free rates contributes positively to returns, and for investment grade it continues to dominate the slight drag We do not expect meaningful changes in the structural demand for on returns arising from tight IG spread levels. The opposite is true ultra-long dated Gilts. Over our forecast horizon, UK pension and for high yield, where the boost from the risk-free rate is much insurance companies represent a significant source of demand for long-dated UK paper; however, in the post-global financial crisis smaller than the drag on returns coming from the tight prevailing (GFC) period, this demand has diminished. The result is two distinct credit spread. regimes for the UK long end since the mid 1990s: In the pre-GFC Our estimates of equilibrium credit spread for U.S. investment era, an inverted 10s30s curve was commonplace (averaging -20bps grade are unchanged from last year, with 10-year IG spreads at between 1996 and 2008), and in the post-GFC period the 10s30s 150bps and 30-year IG spreads at 175bps. Combining the spread curve was, on average, 80bps positively sloped. We believe the and duration components leaves us with a return estimate of former regime will eventually reassert itself as the management of 3.50% for 10-year U.S. IG corporate credit, a pickup of 25bps over inflation risk and fund de-risking create ongoing demand for ultra- last year’s estimate. By contrast, U.S. HY returns drop 50bps from long dated Gilts—in both inflation-linked and nominal bonds. We 5.75% to 5.25%. thus expect the UK 10s30s curve to be flat in equilibrium, with both 10-year and 30-year Gilts yielding 3.00%. This leaves the UK with In last year’s edition, we examined the patterns of default and by far the flattest yield curve of the developed nations, with an recovery rates across credit markets and concluded that both were inversion of the long end of the curve a significant risk over our remarkably stable in the long term. While discrete credit cycles can forecast horizon. be more or less severe, long-term average default rates are strongly mean-reverting to 3.50% and recovery rates to 40%. We also JAPANESE RATES observed that the average level of U.S. high yield spreads in the last 20 years is meaningfully higher than in the 20 years ended Given Japan’s low level of trend growth and our belief that, immediately before the global financial crisis. While this is on average, inflation rates in Japan will fall well short of the 2% commonly explained away as a “liquidity premium,” we do not target, we expect some form of loose monetary policy throughout believe that long-term patterns of default and recovery justify the our forecast horizon. We therefore reduce our expectation of additional post-GFC spread and thus expect the post-GFC premium equilibrium cash rates by 50bps to just 0.50%, resulting in to gradually accrue back to investors over the longer run. As a negative real cash rates in equilibrium. We assume longer-dated result, we maintain our expectation that over our forecast horizon, yields will normalize in line with global trajectories to equilibrium U.S. HY spreads will, on average, retain a residual 25bps of liquidity levels of 1.25% for 10-year Japanese government bonds (JGBs) and premium,4 leading to an equilibrium spread assumption of 500bps. 1.75% for 30-year JGBs. This implies a steeper curve in Japan than This is wider than prevailing spreads and thus implies a drag on U.S. we’ve seen in recent years, but with real 10-year yields remaining HY returns. Nevertheless, compared with other fixed income assets, below equilibrium nominal GDP. We infer that the Bank of Japan such a level of return remains attractive (Exhibit 7). (BoJ) will need to balance a steeper curve (which aids the domestic financial sector) with equilibrium yields below nominal growth rates (which aid debt sustainability). 4 The impact of the liquidity premium is explored in our 2017 Long-Term Capital Market Assumptions, p. 53.

58 LONG-TERM CAPITAL MARKET ASSUMPTIONS G4 GOVERNMENT BONDS: SLOW ROAD, LOW YIELDS

GLOBAL EMERGING MARKET DEBT: CYCLICAL SUPPORT, STRUCTURAL REPAIR, CHINA EMERGENCE

Over the past year, emerging markets have benefited from a modest decline in the U.S. dollar and a broad-based upswing in CHINESE BOND MARKET cyclical growth. Although many EM economies still face a long China’s domestic bond market ranks as the third largest in the road to stabilization, the early signs are encouraging. EM external world, with USD 10.8 trillion (measured by total debt securities debt remains low, and the current account balances of the more outstanding) as of September 30, 2017. The size of the market fragile EM nations5 continue to improve. The EM corporate sector has increased substantially over the last decade, with an has areas of vulnerability but, in general, is of superior credit average annual growth rate in excess of 16%. quality to U.S. high yield, with roughly two-thirds of the CEMBI About 40% of the total debt outstanding—representing broad diversified index made up of investment grade credits. 2,190 securities with a market value of USD 4.5 trillion as of Prevailing spread levels of both sovereign (EMBIG diversified) and September 30, 2017—qualifies for inclusion in the Bloomberg corporate EM debt are currently close to our equilibrium estimates Barclays China Aggregate Index. The index is dominated by government and government-related issuers, including of 325bps and 375bps, respectively. This translates to return agencies. Corporates, primarily from the financial sector, estimates of 6.25% for EM sovereign debt in local currency, represent about 14% of the index. 5.25% for hard currency EM sovereign debt, and 5.25% for EM On May 16, the People’s Bank of China and Hong Kong corporate debt. Monetary Authority jointly announced the approval of the Bond Connect program, which aims to provide a platform for global China’s financial markets will have an increasing presence on the investors to directly access China’s bond market. global stage over the next decade or two, a trend that will bear While details of this program are still under discussion, it close attention both for EM debt considerations and for the broader could help to overcome a key hurdle for the inclusion of the tone across global government debt markets. China bond market in major global bond indices. We believe As Chinese financial markets continue to liberalize, we expect that ownership of Chinese bonds by foreign investors may increased access to China’s USD 10.8 trillion bond market and USD 9 rise considerably in response to this change, given that they trillion equity market.6 We see equilibrium yields of 4.00% for currently own less than 2% of the onshore bond market. Chinese 10-year sovereign bonds, and 4.50% for Chinese 30-year sovereign bonds. (A description of Chinese bond indices is included * We discuss the implications of China’s ongoing financial liberalization for fixed income investments in “The cost of capital in China’s changing markets,” in the sidebar, “Chinese bond market,” and a more detailed paper featured in our 2018 Long-Term Capital Market Assumptions. is available on our website.)

5 Based on the “fragile five” of Brazil, India, Indonesia, South Africa and Turkey. 6 Our estimate of aggregated Chinese equity markets, both on and off-shore.

For U.S. high yield, we make an equilibrium spread assumption of 500bps, which is wider than prevailing spreads and implies a slight drag on U.S. high yield returns EXHIBIT 7: 10-YEAR RETURN ESTIMATES FOR S&P 500, IG, HY; AVERAGE SPREAD BETWEEN THEM SHOWN IN RETURN TERMS U.S. long Treasuries U.S. investment grade corporate bonds U.S. high yield bonds U.S. large cap equity High yield-long Treasuries dierential % 12

0 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 ’17 ’18

Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. IG: investment grade; HY: high yield

J.P. MORGAN ASSET MANAGEMENT 59 EQUITY MARKET ASSUMPTIONS

Modestly but steadily lower returns Patrik Schöwitz, CFA, Global Strategist, Multi-Asset Solutions Stephen Macklow-Smith, Portfolio Manager, European Equity Group Michael Albrecht, CFA, Global Strategist, Multi-Asset Solutions Yann Vestring, CFA, Quantitative Analyst and Portfolio Manager, Multi-Asset Solutions Nika Mosenthal, Analyst, Multi-Asset Solutions

IN BRIEF • Our equity assumptions project lower expected returns in developed and emerging markets. In line with recent years, we see returns over our time horizon falling modestly but steadily behind what historical experience would suggest. The gap between emerging and developed market equity returns narrows somewhat from our last edition in U.S. dollar (USD) terms.

• Another year of strong equity performance has coincided with an expansion of corporate profit margins relative to long-run averages. Across most markets, our return assumptions also continue to face valuation headwinds, notably in the U.S., and anticipated pressure on margins.

• Our view on Japanese equities remains unchanged despite strong performance, as we raise our view of achievable long-term margins and valuations. Despite slow GDP growth, we believe changing corporate behavior in Japan will produce greater capital returns to shareholders. Our euro area assumption falls modestly; our expected U.S. and UK returns fall slightly more.

• Our emerging market (EM) equity return assumptions decline moderately, driven by a fall in emerging Asia return expectations, stemming from our presumption of lower GDP (and thus earnings) growth, as well as higher valuations. In contrast, assumptions for Europe, the Middle East and Africa (EMEA) and Latin America are largely unchanged in aggregate local currency terms, although EMEA USD returns suffer from a foreign exchange downgrade.

• The return of capital to shareholders in the form of dividends and buybacks will be a crucial component of future returns. We continue to forecast that a high proportion of earnings, relative to historical averages, will be distributed.

• We introduce assumptions for global convertible bonds and global credit-sensitive convertible bonds.

60 LONG-TERM CAPITAL MARKET ASSUMPTIONS MODESTLY BUT STEADILY LOWER RETURNS

MODESTLY LOWER, AGAIN

The message from our equity assumptions for the past several The impact is to lift DM returns to 6.00% in USD, though EM years has been one of expected returns falling steadily further returns remain the same 8.00% as in local terms. behind historical experience. This year is no exception. We expect the return of capital to shareholders in the form of Once again, 2017 was a year when most equity indices rose dividends and buybacks to be a crucial component of future returns. substantially in excess of our long-term return estimates and of Our analysis shows that in some markets in recent years more than delivered earnings growth. The result of higher starting margins 100% of earnings were distributed. While we do not expect these and valuations, combined with a number of country-specific extreme payout rates to continue, we forecast that the vast majority macroeconomic downgrades, is that our returns expectations of earnings will be distributed (Exhibit 1). Our framework is agnostic have broadly declined. on whether this behavior arises from a dearth of capital investment opportunities, ultra-low interest rates or a combination of the two. In local currency terms, our expectations for long-term developed Another intriguing possibility is that investors’ demographic profile market (DM) equity returns have fallen to 5.50% from 6.00% last has shifted and they are now more interested in income than in year, while our expectations for emerging market equity returns growth—and with meager levels of income on offer in sovereign have fallen to 8.00% from 8.75%. The expected gap in returns debt markets, equities and credit are substitutes. between emerging and developed equities narrows 25 basis points (bps) from last year. Developments in banking in the U.S. and Europe may also be part of the explanation for changes in the distribution of earnings. U.S. While the magnitude of the downgrade is similar for both blocs, the banks raised a lot of capital very quickly after the financial crisis reasons differ. In developed markets, our downgrade is driven by vs. eurozone banks’ rather more drawn-out process. We believe, more challenges for corporate profit margins and slightly higher though, that eurozone banks have now almost completed their equity valuations in aggregate. In emerging markets, higher capital-raising, and their balance sheets are sufficiently robust to valuations and lower GDP (and thus earnings) growth expectations allow for greater distribution of dividends. In the U.S., the results account for more of the downgrade. But on aggregate, it is of recent stress tests suggest that distributions from banks should predominantly cyclical, not secular. also increase. In the medium term, the much higher level of Tier 1 We continue to expect currency movements to be a major driver capital in the banking system, along with other forms of hybrid of returns for unhedged equity investors. In particular, we expect capital, implies that capital buffers are much more robust. the USD to weaken moderately over our assumption horizon, This likely means that in the next growth downturn banking boosting the attractiveness of international equity markets systems on both sides of the Atlantic will have less need to raise (emerging and developed) to U.S. dollar-based investors. new equity capital.

Payout ratios are expected to be elevated relative to history across many developed markets EXHIBIT 1: PAYOUT RATIO (DIVIDENDS + BUYBACKS, % OF EARNINGS), HISTORICAL AND PROJECTION

1995-2006 Last 10 years, 2007-16 Projection, next 10-15 years % of earnings 100

50 U.S. large cap Euro area U.K. Japan

Source: Thomson Reuters Datastream, Citigroup, J.P. Morgan Asset Management; data as of August 7, 2017.

J.P. MORGAN ASSET MANAGEMENT 61 EQUITY MARKET ASSUMPTIONS

BUILDING OUR FORECASTS Slower revenue growth and high valuations subdue U.S. large cap returns relative to history We continue to rely on the equity return assumptions EXHIBIT 3: CONTRIBUTION TO TOTAL RETURNS, % ANNUAL, FOR U.S. LARGE methodology we introduced in our 2015 assumptions, CAP EQUITIES summarized in Exhibit 2. Dividend yield Buybacks* Revenue growth Margins impact Gross dilution Valuation impact Total return % Our equity assumptions methodology breaks equity 24

returns into easy-to-forecast return drivers 18

EXHIBIT 2: BUILDING BLOCKS–ANATOMY OF EQUITY TOTAL 12 RETURNS 1. Aggregate revenue growth 6 2. × Aggregate earnings growth / revenue growth (margins) 0 = Aggregate earnings growth -6 3. × Earnings per share (EPS) growth / aggregate earnings -12 growth (net dilution) = EPS growth 1990-99 2000-09 2010-16 LTCMA ’18

4. × Price return / EPS growth (valuations) = Price return Source: Thomson Reuters Datastream, U.S. Bureau of Economic Analysis, Citigroup, J.P. Morgan Asset Management; data as of September 30, 2017. 5. + Dividends (carry) = Total return * Buybacks are included in “gross dilution” before 2000, i.e., as net dilution, due to limited data availability.

Similar to DuPont analysis, this methodology allows us to In the euro area, we moderately decrease our expectations for decompose total returns structurally into easily forecastable ratios future returns of large cap equities to 5.75% from 6.00%, reflecting as drivers of returns. It enables us to account explicitly for the much wider starting profit margins, offset to some degree by global composition of corporate revenues—and how fast different higher revenue growth as we raise our expectations for nominal regions are growing—as well as the normalization of profit margins euro area GDP, and for a higher level of distribution to and valuations, and the impact of share buybacks and dilution. shareholders. In recent years, the financial sector has needed to Perhaps most important, it ties together complex interrelationships raise a lot of capital. We believe this process is largely complete among these factors to ensure that retained earnings and gross and therefore, the net effect will be to raise payouts. European dilution imply a future book value that is consistent with projected companies’ valuations have risen since last year, and the return to return on equity and future earnings. This framework—analogous to earnings growth means that margins have already recovered to Robert Higgins’ sustainable growth rate (SGR) concept—suggests where they are now a roughly neutral factor for future returns. that higher shareholder payouts, for instance, would come at the That stands in contrast to U.S. equities, which only during the late expense of slower earnings growth, other things being equal. 1990s tech boom were more expensive than they are today, relative to earnings. Europe has seen higher levels of valuation five DEVELOPED MARKETS EQUITY RETURN times in the last 40 years. ASSUMPTIONS Japanese equities continue to face a headwind from slow domestic In the U.S., our expected return for large cap equities falls to GDP growth, but the signs of a change in corporate behavior that 5.50% from 6.25%, due to higher starting valuations and greater we have acknowledged in the past remain clear—with greater headwinds from margin normalization (Exhibit 3). Earnings-based capital returns to shareholders—and we believe that this will valuations for the U.S. now sit comfortably above their long-term continue. The prolonged de-rating of Japanese equities from their average, so our conclusion that valuation is a headwind should not heady 1980s peaks is now over, and we expect earnings-based be controversial. Our work on total distribution to shareholders, valuation in the future to be closer to what we see in the rest of the world. Although Japanese equities have delivered strong including both dividends and share buybacks, shows that U.S. performance since last year, valuations are little changed as public companies have been distributing more than 100% of margins have continued to grow since a surge induced by the yen’s earnings in recent years. We believe that this is the product of the weakness in 2013–14—one of many signs of a positive change in recent ultra-low interest rate environment and that as interest corporate governance. Accordingly, we have again raised our rates normalize, distribution will decline as a percentage of assumptions for long-run sustainable Japanese valuations and earnings but remain high. margins, toward levels closer to those in Europe.

62 LONG-TERM CAPITAL MARKET ASSUMPTIONS MODESTLY BUT STEADILY LOWER RETURNS

The UK market has also performed well since last year. But the Foreign revenue shares vary considerably across major data as yet shows no earnings recovery from recent years’ developed markets downtrend caused by falling commodity prices and financial sector EXHIBIT 4: INTERNATIONAL REVENUE BREAKDOWN FOR G4 MARKETS woes. Although the outlook for the UK in the wake of Brexit U.S. Europe UK Japan Asia Paciﬁc ex-Japan EM/other remains uncertain, the vast majority of the UK equity market is % internationally, rather than domestically, focused (Exhibit 4). This 100 is another reason our framework continues to assume a recovery 80 in earnings from these depressed levels, even though future profitability will likely remain impaired relative to history, given 60 lower commodity prices. A strengthening currency is another potential margin headwind, although, conversely, reported 40 corporate earnings have so far offered little evidence that sterling’s fall has provided a tailwind for earnings (though this may 20 just be a reporting lag). Furthermore, rising margins have

0 U.S. EUROZONE UK JAPAN historically often gone hand in hand with stronger sterling in the UK. Overall, this combination of factors means that our expected Source: Thomson Reuters Datastream, J.P. Morgan Asset Management; data as of August 7, 2017. return falls moderately, to 5.50% from 6.25% in last year’s report.

Our equity assumptions call for modestly lower expected returns, in line with the pattern of recent years EXHIBIT 5: SELECTED DEVELOPED MARKET EQUITY RETURN ASSUMPTIONS AND BUILDING BLOCKS Equity assumptions U.S. large cap Euro area UK Japan Revenue growth 5.3 4.6 4.5 3.0 + Margins impact -0.7 -0.3 0.8 -0.7 Earnings growth 4.5 4.3 5.4 2.3 + Gross dilution -2.0 -2.0 -2.0 -2.0 + Buybacks 2.3 1.6 0.7 2.1 EPS growth 4.8 3.8 4.1 2.3 + Valuation impact -1.4 -1.1 -2.1 -0.2 Price return 3.3 2.7 1.9 2.1 + Dividend yield (DY) 2.0 3.0 3.5 2.5 Total return, local currency 5.50% 5.75% 5.50% 4.75% Change vs. 2017 -75bps -25bps -75bps – Source: J.P. Morgan Asset Management; estimates are as of September 30, 2017.

EXHIBIT 6: SELECTED EMERGING MARKET EQUITY RETURN ASSUMPTIONS AND BUILDING BLOCKS Equity assumptions China (H) Korea Taiwan India South Africa Brazil Revenue growth 8.8 6.3 5.1 13.1 9.6 9.6 + Margins impact -0.3 -1.2 -0.5 -0.1 -0.4 0.8 Earnings growth 8.4 5.1 4.6 13.0 9.1 10.5 + Gross dilution -3.0 -1.1 -1.0 -2.8 -1.9 -4.3 + Buybacks 0.3 1.3 0.5 0.3 0.5 0.8 EPS growth 5.4 5.3 4.0 10.1 7.6 6.5 + Valuation impact -0.6 0.7 0.6 -2.8 -2.1 -3.3 Price return 4.8 6.1 4.6 7.0 5.3 3.0 + Dividend yield (DY) 2.8 1.5 3.5 1.5 3.0 3.5 Total return, local currency 7.75% 7.75% 8.25% 8.75% 8.50% 6.75% Change vs. 2017 -200bps -25bps -75bps – +50bps -75bps

Source: J.P. Morgan Asset Management; estimates are as of September 30, 2017.

J.P. MORGAN ASSET MANAGEMENT 63 EQUITY MARKET ASSUMPTIONS

EMERGING MARKET EQUITY RETURN ASSUMPTIONS In Exhibits 5 and 6, we present the building blocks that form the foundation of our DM and EM equity assumptions. Emerging markets should continue to offer a higher return than developed markets. We expect the gap to developed equities to narrow modestly to 250bps in local currency, as noted above, CONVERTIBLE BONDS mostly driven by higher revenue (and thus earnings) growth. For the first time in our LTCMAs’ 22-year history, we are adding From a structural perspective, this higher growth comes courtesy long-term assumptions for convertible bonds. Although this asset of still-high productivity and the potential for a technology catch- class is new to our publication, our multi-asset team has been up, as well as emerging economies’ generally better demographic running convertible bond strategies since 1995. picture vs. developed nations, (with the exception of China). Cyclically, many emerging economies are now recovering from the Convertible bonds—corporate debt securities that provide the downturn triggered by the collapse in commodity prices in 2015– holder with an option to convert into the issuer’s stock at a 16, and recent strong performance across EM equity markets predetermined price—have historically offered investors equity- suggests the long period of underperformance vs. developed like returns with lower volatility and downside protection through markets has likely drawn to a close. Our expectation that the U.S. a fixed income floor. Since Thomson Reuters Global Hedged dollar will weaken over our forecast period also means that, in the Convertible Bond Index started in 1994, through the third quarter medium term, funding pressures for emerging market sovereign of 2017, it has outperformed the MSCI World Index. Convertibles borrowers will dissipate. are suitable for income-oriented strategies due to their fixed income characteristics and can improve the risk-adjusted returns We derive our EM equity assumption by applying the same of balanced stock-bond portfolios due to their asymmetric return methodology to nine large emerging markets and aggregating by profile and diversification benefits. However, like high yield bonds, market capitalization weight. The countries we include account for convertibles have been susceptible to liquidity constraints during more than 85% of the market capitalization in the MSCI Emerging periods of market stress. Markets Index. We once again caution that data history in emerging economies is generally shorter, and data quality less robust, so our Our methodology for calculating convertible bond returns confidence in the resulting assumptions is by nature somewhat accounts for convertibles’ similarities to and differences from lower than for developed markets. Despite this reservation, and the traditional equity and fixed income, as well as the composition of variety of cyclical and structural crosscurrents moving through the convertible indices. While the geographic composition of the emerging market universe, we identify a few common themes. global convertibles universe is similar to that of the MSCI World, it has historically been biased toward smaller companies and cyclical We see a shift among regions: Our local currency assumption for sectors. We incorporate into our convertible bond assumptions the dominant EM Asia region drops to 8.00% (8.25% in USD) from our existing LTCMA numbers for equity and fixed income, along 9.00% (9.75% in USD), driven mostly by lower GDP growth and with convertibles’ equity sensitivity, credit quality, option premium higher starting valuations. In contrast, assumptions fall more and the underlying stocks’ unique characteristics. modestly for EMEA, down only 25bps to 8.50% (7.75% in USD), and in Latin America remain steady at 7.25% (6.75% in USD). This year, our global convertible bond and global credit-sensitive convertible bond assumptions (hedged into USD) The one percentage point decline in EM Asia returns can be are 5.00% and 4.25%, respectively. Credit-sensitive convertibles are traced to declines in all three of the region’s major markets. The securities whose underlying stock trades significantly below the return for Chinese equities falls to 7.75% from 9.75%, owing to conversion price, causing them to behave more like debt than substantially higher starting valuations and lower expected GDP equity. For context, we forecast 6.00% for MSCI AC (All Country) growth. Our Korea return expectation falls to 7.75% from 8.00%, World and 3.25% for global credit returns. Our assumptions for also driven by higher starting valuations and lower GDP growth, convertible bonds (applying our methodology to last year’s data) although partly offset by an expected recovery in margins and have fallen approximately 50bps this year vs. 2017. This change is buybacks. Our Taiwan return assumption drops to 8.25% from consistent with the shifts in our equity and fixed income 9.00% due to lower GDP growth and an expected fall in margins, assumptions, and our outlook for a lower return world over the next partly offset by cheaper valuations vs. last year. Within the 10 to 15 years. relatively stable LatAm and EMEA return expectations, Brazil drops to 6.75% from 7.50%, mostly due to much higher valuations after a year of strong performance. The return assumption for South Africa rises to 8.50% from 8.00% due to a higher GDP growth assumption, slightly offset by more elevated starting margins. All assumptions are in local FX.

64 LONG-TERM CAPITAL MARKET ASSUMPTIONS ALTERNATIVE STRATEGY ASSUMPTIONS

A stable to improving alpha outlook offsets the beta drag Anthony Werley, Chief Portfolio Strategist, Endowments and Foundations Group

IN BRIEF • We see an improving outlook for alternative strategies relative to public markets—largely due to industry and investment trends expected to favor alpha generation and help offset some of the drag from reduced public market expectations. Changes to our alternative return estimates vs. 2017 run from negative to positive, reflecting the unique fundamentals of each strategy type. As always, manager selection will be a critical determinant of investment success across all alternative strategy classes. • Private equity (PE): Reduced public equity assumptions, historically high PE purchase price multiples and significant stores of dry powder lead to a somewhat more pronounced decline in return estimates for PE vs. other alternative strategy classes. A broader opportunity set may help enhance investors’ returns. • Direct lending: Projected returns are up slightly and imply yields meaningfully higher than comparable public market credits. Premiums can be expected to fall as this private market matures. • Hedge funds: We mark up return estimates for most strategies, largely reflecting expectations of a transition from a macro-driven to a more fundamentally driven market environment, supportive of alpha generation. • Real estate: Though return estimates are down slightly, the outlook for real estate remains a relative bright spot. Positive factors include disciplined supply, constrained leverage and the globalization of real estate investment flows. REIT returns are based primarily on regional core real asset return assumptions and incorporate leverage. • Infrastructure: The return outlook for infrastructure equity is unchanged, given two opposing forces: a strong bid pushing up project pricing, and investors’ expectations of higher illiquidity premiums. Infrastructure debt returns are also unchanged, as a lack of investor experience in these markets is expected to keep illiquidity premiums high. • Commodities: Most current indicators (with the exception of those for the energy sector) continue to point to an upturn in the broad commodity cycle. Our return assumptions imply a 150 basis point (bps) real return over inflation, unchanged from last year. For gold, we anticipate a 25bps premium to commodities.

J.P. MORGAN ASSET MANAGEMENT 65 ALTERNATIVE STRATEGY ASSUMPTIONS

OVERVIEW Alternative return assumptions are mostly flat to slightly down, but improved relative to public market returns The improving return outlook for alternative strategies vs. public EXHIBIT 1: SELECTED ALTERNATIVE STRATEGIES—RETURN ASSUMPTIONS (%) markets results from an upgrade in the investment environment 2018 2017 for hedge funds, an incrementally expanded opportunity set in * private equity and hedge funds, and investor discipline in core real PRIVATE EQUITY (USD) 7.25 8.00 estate, infrastructure and private debt. All three trends were U.S. private equity – small cap 6.50 7.50 previously identified in our 2017 alternatives outlook, but in our U.S. private equity – mid cap 6.75 7.75 2018 assumptions the trade-offs and benefits of private vs. public U.S. private equity – large/mega cap 7.50 8.00 market execution options move a little closer to the conditions PRIVATE DEBT (USD) that prevailed prior to the current market cycle. As has been the Direct lending 7.00 6.75 case historically, the return dispersion among managers across HEDGE FUNDS (USD) the alternative investment spectrum remains wide; full Equity long bias 4.75 4.50 compensation for the risks taken in alternative investing lies in Event-driven 4.75 4.75 execution above the average manager’s performance. Relative value 4.50 4.25 Despite a markdown of public equity and credit expectations, Macro 3.75 4.00 alternative return expectations are generally flat to slightly down Diversified** 4.25 3.50 (except for direct lending and most hedge fund strategies, which Conservative† 3.75 3.00 are slightly up, and small and midsize private equity, which are REAL ESTATE - DIRECT (UNLEVERED, LOCAL CURRENCY) down somewhat more significantly) relative to last year’s U.S. core 5.25 5.50 assumptions (Exhibit 1). A clear exposure to non-U.S. markets— U.S. value-added 6.50 7.00 particularly Asia ex-Japan, where we have higher return European ex-UK prime 4.75 5.00 expectations than for U.S. markets—benefits the private equity and European ex-UK non-prime 6.50 7.00 hedge fund assumptions. Likewise, as a more fundamentally driven UK core 4.75 5.25 and less central bank/macro market-driven environment continues Asia Pacific core 5.50 5.50 to evolve, we would expect hedge funds to be in a clearly better †† operating environment. Interestingly, while public market REITS (LEVERED, LOCAL CURRENCY) valuations and returns have been bid up over the past year, U.S. REITs 6.25 investors have asked for higher upfront compensation in the form European REITs 7.00 of higher starting yields in core real estate, infrastructure equity Asia Pacific REITs 7.00 and private debt. Global REITs 6.50 GLOBAL INFRASTRUCTURE (USD) The coast is still far from all-clear as asset size builds in private Equity – direct 6.25 6.25 equity even as purchase price multiples reach or exceed previous peaks. Hedge funds face competition from smart beta and Debt 4.25 4.25 quantitative liquid market operators. U.S. real estate core supply COMMODITIES (USD) 3.75 3.75 is subdued but increasing, and the infrastructure regulatory Gold 4.00 4.00 outlook still has a bias toward modest allowed rates of return not Source: J.P. Morgan Asset Management; estimates as of September 30, 2017, and attractive enough to induce new capital projects. All things September 30, 2016. * The private equity composite is AUM-weighted: 60% large cap and mega cap, 30% mid considered, the risk-adjusted outlook for illiquid investing has cap and 10% small cap. Capitalization size categories refer to the size of the asset pool, improved on the margin, at least vs. public market options. which has a direct correlation to the size of companies acquired, except in the case of mega cap. ** The diversified assumption now represents the projected return for multi-strategy hedge funds (vs. funds of funds, as in previous LTCMAs). † “Conservative” represents the projected return for multi-strategy hedge funds that seek to achieve consistent returns and low overall portfolio volatility by primarily investing in lower volatility strategies such as equity market neutral and fixed income arbitrage. †† 2018 assumptions for REITs are levered; in previous LTCMAs, these assumptions (not shown) were unlevered.

66 LONG-TERM CAPITAL MARKET ASSUMPTIONS A STABLE TO IMPROVING ALPHA OUTLOOK OFFSETS THE BETA DRAG

PRIVATE EQUITY Manager choice remains a key determinant of successful PE investing, particularly when investing in smaller funds The principal driver of our private equity average return EXHIBIT 3: HISTORICAL PRIVATE EQUITY DISPERSION BY SIZE OF estimates—public equity return assumptions—has been reduced. FUND,* IRR OF VINTAGE YEARS (2005-15) (%) This reduction, together with historically high PE purchase price 25th percentile 50th percentile 75th percentile multiples and a large store of assets to be invested, leads us to % 20 reduce our long-term PE return estimates (Exhibit 2). There are, 18.1 16.0 however, some positive developments within the industry. 15 14.8 Investment discipline has extended into new, higher growth and less efficient markets, and there is greater opportunity to invest in 10 9.8 10.0 9.7 a more concentrated, less fee-burdened “co-invest” option. 5 5.1

Private equity can still offer the potential to earn a premium over 2.2 public equity returns, but expectations of premia equivalent to 0 -0.1 those traditionally associated with private equity are unlikely to be -5 met. As we have said in the past, fair compensation for private <$500mn $500mn–$2bn >$2bn equity’s additional illiquidity, leverage and overall investment model risk accrues to those who can move up the internal rate of Source: Burgiss Private iQ, J.P. Morgan Asset Management; data as of December 2016. *Includes buyout and expansion capital funds. return (IRR) dispersion of PE funds through skillful due diligence and manager access; at the average manager level, expectations are for a very modest premium to public markets (Exhibit 3). Current market conditions

Along many dimensions, the private equity strategy class currently Private equity return estimates are down, given lower public resembles most of the capital markets: generous valuations, a equity assumptions and challenging PE industry conditions large volume of assets to be invested and an extension along the EXHIBIT 2: PRIVATE EQUITY RETURN ASSUMPTIONS AND EQUITY BETA BUILDING BLOCKS risk curve in pursuit of potentially improved returns vs. the traditional portfolio profile. Valuations as measured by purchase price multiples have risen meaningfully since the bottom of the last market cycle and are now at or above the multiples paid at the prior cycle peak (Exhibit 4)—not just in the U.S. but generally across the developed markets. As one measure of risk-taking,

Small PE (<$500mn) Mid PE ($500mn-$2bn) Large/mega PE (>$2bn) Cap-weighted* however, the average debt multiples of issuers are still somewhat EQUITY BETAS below those at the top of the last cycle (Exhibit 5). Small cap -0.19 Regardless of the industry environment, private equity is seen as Mid cap 0.63 0.63 0.27 an antidote to lower expected public equity returns and, as such, Europe 0.21 portfolio allocations to private equity have risen precipitously over Asia ex-Japan 0.36 the past few years. A strong distribution environment in recent Adjusted R2 0.38 0.57 0.70 years is also contributing to the capital available for investment. INTERNAL RATE OF RETURN ASSUMPTIONS (USD, %) Fund sponsors have responded in kind by raising ever-larger funds 2018 6.50 6.75 7.50 7.25 (Exhibit 6). Anecdotally, the largest fund size so far this cycle has 2017 7.50 7.75 8.00 8.00 surpassed the peak fund size of 2007 by approximately 30%. Source: J.P. Morgan Asset Management; regression data from June 30, 2006, Additional pressure on returns comes from the well-heeled to December 31, 2016; estimates as of September 30, 2016 and September 30, 2017. competition of sovereign wealth funds and large pension plans * The private equity composite is AUM-weighted: 60% large cap and mega cap, 30% mid cap and 10% small cap. Capitalization size categories refer to the size of the looking to enhance returns by going direct and avoiding fund asset pool, which has a direct correlation to the size of companies acquired, except sponsor fees. Unlike other alternative strategies where historically in the case of mega cap. higher returns have been challenged, there is no expectation of fee abatement within the illiquid equity space.

J.P. MORGAN ASSET MANAGEMENT 67 ALTERNATIVE STRATEGY ASSUMPTIONS

Purchase price multiples exceed those at the peak of the last cycle ... EXHIBIT 4: PRIVATE EQUITY PURCHASE PRICE MULTIPLES Purchase price Fees/expenses 12x

10x

0x ’00 ’01 ’02 ’03 ’04 ’05 ’06 ’07 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 1H ’17 (116) (51) (40) (66) (127) (134) (178) (’7) (69) (23) (78) (97) (97) (95) (136) (114) (105) (80)

Source: S&P Capital IQ, J.P. Morgan Asset Management; data as of June 30, 2017. Number of deals shown in parentheses.

… while debt multiples are still below the prior cycle’s peak EXHIBIT 5: AVERAGE DEBT MULTIPLE OF ISSUERS (Ebitda > $50M)

Subordinated debt/Ebitda Other senior debt/Ebitda Second lien debt/Ebitda First lien debt/Ebitda 7x

0x 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 1H17

Source: S&P Capital IQ, J.P. Morgan Asset Management; data as of June 30, 2017.

Fundraising, driven by increasing private equity allocations, has risen dramatically EXHIBIT 6: HISTORICAL PRIVATE EQUITY FUNDRAISING BY YEAR

Number of funds (LHS) Aggregate capital raised ($bn) 1,400 450 400 1,200 350 1,000 300 800 250

600 200 150 400 100 200 50

0 0 ’00 ’01 ’02 ’03 ’04 ’05 ’06 ’07 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 1H17

Source: Preqin, J.P. Morgan Asset Management; data as of June 30, 2017.

68 LONG-TERM CAPITAL MARKET ASSUMPTIONS A STABLE TO IMPROVING ALPHA OUTLOOK OFFSETS THE BETA DRAG

New opportunities and extensions along the risk curve The outlook for alpha

Several developments have the potential to enhance private Full purchase price multiples, a sizable store of dry powder, equity returns, including an expansion of the PE opportunity set increased competition and no discernible return advantage from and new options for accessing PE investments. co-invests, on average, argue for less of a premium above public markets across the fund capitalization spectrum. Our long-term Within large/mega and midsize funds, international market return assumptions are marked down accordingly from last exposures, specifically in Europe, have appeared for a number year’s, given the lower underlying beta expectation and steady of years. And, for the first time, our 2018 equity beta statistics premium over public markets. A marginal exception is made in record a measurable Asia ex-Japan exposure in large/mega funds the large/mega capitalization fund segment to reflect the (see Exhibit 2). Such exposures offer the base case of higher increased internationalization of portfolio holdings. This is in line underlying public market returns vs. those of more U.S. with our expectation of higher underlying public market returns, domestically focused funds. especially in Asia, and the potential, at least, for improved alpha Within the growth equity subcomponent of private equity, generation vs. the available alpha in U.S. markets. the higher growth sectors of technology and life sciences offer Finally, it cannot be said too often: Our assumptions represent opportunities for return above the outlook for general GDP-oriented average manager returns. The true return-enhancing potential investing. Specifically, cybersecurity, artificial intelligence, disruptive of private equity, as demonstrated by top quartile returns, is technology, industrial innovation and millennial-focused products realized through careful due diligence and selection of top- 1 and services portend better growth and potential return outlooks. performing managers. The most notable development within the past few years is the increasing role of co-investment opportunities—that is, single- company, often reduced-fee investment options—within a diversified private equity portfolio. Clearly, there is potential for higher overall portfolio returns when allocating to fewer, more select, higher conviction opportunities, coupled with a fee reduction. Yet our research indicates that, on average, actual results have not lived up to that premise. These concentrated bets have produced returns in line with, or slightly below, overall portfolio performance. Even so, co-invests have served to put some dry powder to work and have reduced the drag from fees at the margin. As fund sizes increase, however, co-investment opportunities are likely to become less available.

Co-investments serve, once again, to highlight the role that manager and investment selection, not allocation, play in generating excess returns as fund sponsors and investors go farther out on the risk curve through more concentrated portfolios.

1 See “The impact of technology on long-term potential economic growth,” J.P. Morgan Asset Management, 2018 Long-Term Capital Market Assumptions for more on this topic.

J.P. MORGAN ASSET MANAGEMENT 69 ALTERNATIVE STRATEGY ASSUMPTIONS

DIRECT LENDING

Our 2018 long-term return estimate for direct lending is 7.00%, up Composite characteristics 25 basis points from last year’s estimate (Exhibit 7). This increase is due largely to a broadening of the universe we model to Direct lending characteristics are modeled primarily on the incorporate somewhat higher quality borrowers, resulting in a Cliffwater Direct Lending Index (CDLI), a composite of approximately smaller downward adjustment for expected credit losses. $90 billion of unrated floating rate loans. In developing our assumptions, we have modified the CDLI characteristics to include a Direct lending/private debt volumes continue to rise strongly, slightly broader, higher quality universe, reflecting our assessment reflecting the benefits to borrowers of speed and certainty of of future asset characteristics as this market matures. The assumed execution, a single counterparty and the flexibility of debt debt structure of the enterprise modeled is weighted approximately structures that working with a customized provider affords. 65% to first lien and senior secured loans and 35% to second lien At the same time, regulatory constraints on bank lending are still and junior securitized loans. For the majority of enterprises, Ebitda in force for the traditional providers of credit to this middle falls in the $50 million to $100 million range. Loans are generally a market corporate segment. Current investors receive a significant five-year maturity, though through prepayments and refinancing the illiquidity premium for taking on unrated, smaller size loans with a effective loan life is three years.2 The current starting terms for the credit profile that is similar to leveraged loans but with a more CDLI composite are a 9.25% cash yield and 75bps of original issue cyclical bias. As assets and marginal operators are attracted to the discount (OID) and other concessions, with a 1% net credit loss. Our space, we anticipate a degradation of both the illiquidity premium forward assumption suggests a meaningful, though smaller, received by investors and the historically strong underwriting premium to comparable public market credit options—one we view results experienced by the industry. As direct lending assets grow as more likely to prevail in an excess return, capital-attracting and become a more institutionalized portfolio allocation, environment like that of the maturing direct lending market. the excess premium to publicly available credit should fall, consistent with the experience of other public-to-private management dynamics, such as private equity.

Direct lending is expected to offer yields meaningfully above comparable public market credits, even as this private market matures EXHIBIT 7: DIRECT LENDING—RETURN ASSUMPTIONS AND BUILDING BLOCKS (USD, %)

BUILDING BLOCKS 2018 2017 Cash yield (interest income/coupon) 8.00 8.00 OID, prepayment fees, other concessions 0.50 0.50 Credit loss net -1.50 -1.75 Total return 7.00 6.75

Source: J.P. Morgan Asset Management; estimates as of September 30, 2016 and September 30, 2017.

2 Underlying the Cliffwater Direct Lending Index are unrated floating rate loans, selectively extended to middle market companies. The index is weighted approximately 50% to first lien and senior secured loans, 25% to second lien and junior secured loans and 25% to equity, mezzanine and other junior securities. Loans generally have a five-year maturity, reduced through prepayments and refinancing to an average loan life of three years.

70 LONG-TERM CAPITAL MARKET ASSUMPTIONS A STABLE TO IMPROVING ALPHA OUTLOOK OFFSETS THE BETA DRAG

HEDGE FUNDS A more fundamentally driven market is expected to boost returns for most hedge fund strategy types Our 2018 long-term return assumptions for hedge funds are EXHIBIT 8: HEDGE FUND RETURN ASSUMPTIONS (USD, %) marked higher vs. our 2017 assumptions for all strategies except event-driven (which is flat) and macro (which is down slightly), 2018 2017 even as our beta assumptions across the capital markets are Equity long bias 4.75 4.50 lowered (Exhibit 8). These marginally higher returns assume Event-driven 4.75 4.75 improvements in investment and industry conditions but are Relative value 4.50 4.25 based primarily on the premise that the return-corrosive, central Macro 3.75 4.00 bank-driven, massive liquidity conditions of the past several years Diversified* 4.25 3.50 will give way to a fundamentally driven market environment—one Conservative** 3.75 3.00 more conducive to alpha generation. Source: J.P. Morgan Asset Management; estimates as of September 30, 2016 and September 30, 2017. Projecting multi-strategy vs. fund of funds returns * The diversified assumption now represents the projected return for multi-strategy hedge funds (vs. funds of funds as in previous LTCMAs). ** “Conservative” represents Our definition of diversified hedge funds has changed this year; the projected return for multi-strategy hedge funds that seek to achieve consistent return assumptions now represent multi-strategy funds vs. funds returns and low overall portfolio volatility by primarily investing in lower volatility of funds. The increase in diversified hedge fund return estimates strategies such as equity market neutral and fixed income arbitrage. for 2018 vs. 2017 is due largely to the elimination of funds of funds’ additional fee level and the higher expected return Potential for an alpha upturn potential of the more flexible multi-strategy fund structure. Since the Great Recession, the trend for alpha (residual return not described by beta) has been clearly negative, with the exception of Improving industry conditions equity long bias strategies (Exhibit 9). These trends have been Many of the conditions that depressed hedge fund returns prior to consistent with the macro-driven market conditions of the period, the Great Recession era are still operative—in particular, too large characterized by declining interest rates (and an accompanying an asset base chasing a diminished opportunity set and decline in hedge fund net interest income), rising return competition from non-traditional approaches such as smart beta correlations, low market volatility and massive liquidity throughout and other quantitative strategies. On the positive side, changing the financial system (which, among other effects, has enabled the fee structures and a winnowing-out of the bottom tier of survivorship of otherwise less viable entities, hurting the managers owing to waning asset flows and a deterioration of the performance of the short book in more fundamentally oriented hedge fund business model should help increase returns. strategies). Macro—specifically, systematic macro strategies—face a more problematic outlook because the powerful bull market in fixed income has been a key generator of historical macro returns.

We see a bottoming and gradual upturn of negative alpha trends as the industry and market environment improve EXHIBIT 9: TREND LINES FOR 36-MONTH ROLLING ALPHAS* BY MAJOR HEDGE FUND STRATEGY CLASS EQUITY LONG BIAS EVENT-DRIVEN % % 1.0 3.0 2.0 0.0 1.0 0.0 -1.0 -1.0 -2.0 -2.0 -3.0 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 ’17 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 ’17 MACRO RELATIVE VALUE % % 6.0 2.0 4.0 1.0 2.0 0.0 0.0 -2.0 -1.0 -4.0 -6.0 -2.0 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 ’17 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 ’17

Source: Hedge Fund Research, Bloomberg, J.P. Morgan Asset Management for estimates; data as of May 31, 2017. *Alpha is defined here as the difference between actual composite returns and estimated core beta returns for a given hedge fund strategy. Core beta returns are estimated using J.P. Morgan Asset Management proprietary regression models and actual historical values for the traditional asset class/market drivers of return.

J.P. MORGAN ASSET MANAGEMENT 71 ALTERNATIVE STRATEGY ASSUMPTIONS

With interest rates at an inflection point, this powerful source of Manager selection is critical in realizing the investment potential of return is, at a minimum, diminished going forward. hedge funds Manager selection matters EXHIBIT 10: DISPERSION OF MANAGER RETURNS (%), JULY 2012-JUNE 2017 25th percentile 50th percentile 75th percentile As explained in “Building blocks of hedge fund return estimation % (average industry conditions),” our approach to projecting 12 9.7 average hedge fund industry returns incorporates three basic 9.2 components—core beta returns, alpha trends and alpha potential— 8 7.8 but these are average return assumptions. 5.8 5.7 5.2 4 3.6 While beta continues to be the overwhelming driver of returns at 2.7 3.0 2.7 the average manager level, hedge funds exhibit a high alpha-to- 0.6 beta ratio vs. traditional asset managers as you move upward 0 through the peer group performance ranking. Likewise, the wide -2.0 differential in results between top vs. bottom performers is a key -4 Event-driven Macro Relative value Equity long bias risk of the strategy class. Manager selection remains a critical determinant of the extent to which investors realize the potential of Source: Hedge Fund Research, J.P. Morgan Asset Management; data as of June 30, 2017. hedge funds to generate returns equivalent to a stock-bond combination and to diversify and risk-adjust multi-asset class portfolio performance (Exhibit 10).

BUILDING BLOCKS OF HEDGE FUND RETURN ESTIMATION (AVERAGE INDUSTRY CONDITIONS) CORE BETA RETURNS: the primary component of our projected hedge fund returns—estimated as the product of beta exposures (from our proprietary regression models; see chart below) and our long-term return assumptions for traditional asset classes • Our analysis finds beta exposures increasingly rotating toward higher return non-U.S. markets. ALPHA TRENDS: based on historical alpha trends, adjusted for forward-looking expectations (Exhibit 9) • We expect the long-running negative alpha trend to moderate as fundamentals increasingly drive performance. ALPHA POTENTIAL: further adjustments, based on our interpretation of the impact of industry conditions on the forward-looking alpha potential of each strategy class • We anticipate a fee reduction of at least 25bps at the average manager level, industry-wide.

DERIVED EQUITY BETA EXPOSURES (COEFFICIENTS) AND GOODNESS OF FIT (R2) STATISTICS Long bias Event-driven Relative value Macro Diversified* Conservative** Large cap -0.18 Mid cap 0.30 0.13 Small cap -0.03 -0.02 EAFE 0.15 0.10 0.07 0.19 0.12 Japan 0.05 Asia ex-Japan 0.13 0.05 0.06 Commodities 0.05 0.06 0.06 0.08 0.07 0.07 U.S. aggregate -0.40 -0.37 0.18 U.S. high yield 0.00 0.17 0.23 -0.20 0.03 0.04 World gov. ex-U.S. -0.16 0.02 -0.20 -0.19 EM corporate 0.11 0.05 Adj. R² 0.93 0.83 0.84 0.23 0.68 0.67

Source: Bloomberg, J.P. Morgan Asset Management. The time frame for regression analysis is December 31, 2004, through May 31, 2017. *The diversified assumption now represents the projected return for multi-strategy hedge funds (vs. funds of funds as in previous LTCMAs). **The conservative assumption represents the projected return for multi-strategy hedge funds that seek to achieve consistent returns and low overall portfolio volatility by primarily investing in lower volatility strategies such as equity market neutral and fixed income arbitrage. For further details on our methodology, please see J.P. Morgan Asset Management, 2017 Long-Term Capital Market Assumptions, pp. 65-66.

72 LONG-TERM CAPITAL MARKET ASSUMPTIONS A STABLE TO IMPROVING ALPHA OUTLOOK OFFSETS THE BETA DRAG

REAL ESTATE trends for returns may differ from region to region and city to city, in general, variations are likely to be muted. Over our investment We see a neutral to positive outlook for real estate relative to horizon, changing demographics and online vs. in-store purchasing most other asset classes, varying by real estate sector. We expect will likely be global phenomena impacting traditional patterns of current supply discipline to support returns above levels typical of asset utilization. The supply discipline of the current U.S. and this later stage in the real estate cycle, contributing to this relative European cycles and greater anticipated constraint in developed performance over the long term. Our 2018 Long-Term Capital Asia Pacific economies dominate the long-term outlook for real Market Assumptions (LTCMAs) for real estate are, however, revised estate returns compared with most other investment options, slightly downward from last year’s projections, save for Asia despite the relatively low current absolute cap rates in many parts Pacific core (Exhibit 11). of the industry. Leverage is generally constrained even as net operating incomes (NOIs) remain healthy for this stage of the cycle. The increased globalization of real estate flows may exert a Real estate return expectations are reduced, but still strong relative modest measure of return harmonization across regions, somewhat to most asset classes akin to the dynamics of the global fixed income markets. Within the EXHIBIT 11: REAL ESTATE RETURN ASSUMPTIONS DIRECT (UNLEVERED, LOCAL CURRENCY, %) industry, generally, there exists a wide dispersion of sectoral return outlooks, bounded on the upside by the relatively 2018 2017 controlled development pace and low leverage of core office U.S. core 5.25 5.50 markets and on the downside by the disruption in the brick and U.S. value-added 6.50 7.00 mortar retail segment. Europe ex-UK prime 4.75 5.00 U.S. markets Europe ex-UK non-prime 6.50 7.00 UK core 4.75 5.25 A general tone of caution and regulation in core real estate Asia Pacific core 5.50 5.50 development and lending this cycle continues to define the Source: J.P. Morgan Asset Management; estimates as of September 30, 2016 and attractive relative return expectation in the space vs. most other September 30, 2017. asset class return assumptions. Investors’ shift away from core to value-added has softened pricing and raised core cap rates and Global real estate market trends forward-looking underwriting internal rates of return by about 60bps to 75bps over the past two years (Exhibit 12). Leverage for A number of common themes—the globalization of real estate core assets has remained inexpensive as rates and spreads have investment flows, more modest loan to value vs. past cycles and stayed relatively low even as the aversion to leverage remains historically attractive spreads vs. public market credit—are recurring globally across real estate markets. While the implications of these high. Notably for this late stage in the cycle, NOI growth has outpaced asset appreciation (Exhibit 13).

U.S. core real estate is increasingly attractive vs. comparable public NOI growth is outpacing appreciation for core U.S. real estate credit EXHIBIT 12: U.S. CORE REAL ESTATE IRRs VS. BBB CORPORATE BOND YIELD EXHIBIT 13: FOUR-QUARTER NCREIF ODCE* UNLEVERED PROPERTY TO MATURITY (%) APPRECIATION VS. NOI GROWTH IRRs BBB corporates Appreciation NOI growth % % 8 10

7 6.80 8

6 6 5 4 4 2 3 3.19 1 2 0 ’13 ’14 ’15 ’16 ’17 ’12 ’13 ’14 ’15 ’16 ’17

Source: J.P. Morgan Asset Management; data as of June 30, 2017. Source: National Council of Real Estate Investment Fiduciaries (NCREIF); data as of June *IRRs refer to forward-looking underwriting internal rates of return. 30, 2017. *ODCE is an abbreviation for open end diversified core equity.

J.P. MORGAN ASSET MANAGEMENT 73 ALTERNATIVE STRATEGY ASSUMPTIONS

Over the past six months, a step-up in development activity has approximately 500 weaker malls that may not be able to navigate slightly reduced this cycle’s supply discipline, but at the same the move from “goods distribution” to “placemaking and services time, an upturn in corporate earnings and improved business delivery.”3 outlook may provide further absorption power, as business expectations drive office leasing more than immediate space MULTI-FAMILY: Rents are still at a constructive 3% trend line needs would indicate. However, the recent uptick in IRRs and growth despite significant new development. Value resides in overall tone of the cycle have moderated downside expectations suburban markets even as urban luxury apartment rent growth for the end of this cycle, producing a better compounding effect has recently turned negative. over the term of the projection despite marginally lower current VALUE-ADDED: Flows to value-added strategies are strong at this cap rates vs. long-term equilibrium. stage of the cycle. As this dry powder gets invested, expect a Our basic model for projecting U.S. core returns begins with compression in returns and convergence between core and value- current starting yields, adds estimated growth in yields over the added returns. term and deducts standard industry fees. A final calculation—the Europe ex-UK exit yield adjustment—is made to reflect the impact of a rising or falling yield environment on asset valuation (Exhibit 14). European ex-UK real estate trends are generally in line with the global themes of supply constraint, flow of funds impacting local prices, meaningfully lower leverage vs. the past cycle and an Attractive relative return expectations characterize U.S. core eal estate overall positive tone for the duration of the current cycle. Likewise, European cap rates are by historical standards EXHIBIT 14: U.S. UNLEVERED CORE REAL ESTATE ASSUMPTIONS AND BUILDING BLOCKS (USD, %) expensive, but spreads to interest rates are historically wide and the global yield quest may yet keep returns from reverting to BUILDING BLOCKS long-term equilibriums. Starting yield 5.00 Net cash flow growth 2.50 The bifurcation of markets into prime and non-prime continues, Exit yield adjustment -1.55 with the core/primary cities experiencing better demand and more Standard industry fees -0.70 aggressive pricing than non-prime/tier two cities. As risk appetites increase and the European economy continues its expansion, we IRR assumption 5.25 would expect a catch-up in pricing for the non-prime sector of the Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. market and a compression in the spread between the two tiers.

Sectoral considerations UK

OFFICE: Supply discipline, modest levels of leverage, lower Longer-term projections for pan-European real estate trends financing rates and extended debt maturities characterize this should include a Brexit dynamic that creates a modest return segment, setting up less onerous expectations in the downside of differential between the fortunes of Continental assets and those the cycle. Absorption rates have been slower than expected but in the UK. While UK transaction volumes are down precipitously are likely to improve along with business confidence. since the June 2016 Brexit referendum, returns have been buffered by the arrival of global investors, particularly from China. INDUSTRIAL: The torrid pace of gains, primarily driven by fulfillment London cap rates have risen modestly over the past year, while centers, should moderate, but unmet demand for logistics facilities Continental cap rates are down. The pace and outlook for rent remains robust. We expect the impact of direct-to-consumer growth have slowed more markedly vs. Continental assets. distribution to moderate in the next few years, as the entire logistics However, with healthy fundamentals (including low vacancy levels chain is now focused on reducing inventory overhang. and a somewhat constrained supply in the pipeline), the UK real estate market remains resilient, though a “wait and see” attitude RETAIL: Rumors of the demise of brick and mortar retailers may prevails as the Brexit process unfolds. be exaggerated. Macroeconomic data supports further rationalization of existing facilities rather than more extreme outcomes. Material risk exists within the mall space from

3 “Placemaking” is a multi-faceted approach to the planning, design and management of public spaces, capitalizing on a local community’s assets, inspiration and potential to create public spaces that promote health, happiness and well-being.

74 LONG-TERM CAPITAL MARKET ASSUMPTIONS A STABLE TO IMPROVING ALPHA OUTLOOK OFFSETS THE BETA DRAG

Asia Pacific Real estate investment trusts (REITs)

The 10- to 15-year outlook for APAC real estate has a slightly Publicly traded real estate prices ultimately converge to the healthier tone than that for real estate in the U.S. or European underlying value of the real assets in the REIT index. Our developed markets. This outlook reflects an APAC economic growth methodology for determining a regional REIT return assumption rate that, while structurally slowing, is still anticipated to exceed the starts with that region’s core real asset return assumption. The rate of growth for these developed markets. Slower but still regional core assumption is then adjusted for three factors, increasing urbanization, changing demographics and a rising middle measured at the regional level, specifically impacting REIT returns: income population are expected to drive this APAC growth • net REIT leverage advantage. A stronger web of intra-APAC economic linkages, especially with China, and fewer linkages with a slower-growing, • discount or premium to NAV amortized to its developed U.S. economy should help sustain growth. historical average

Consistent with Asian growth cyclically reaccelerating later than • sectoral differences in the composition of the regional U.S. economic growth, APAC markets, while quite diversified, are REIT index vs. the core real estate composite likely still in the mid-cycle stage. Commercial real estate supply is Given that our methodology uses regional core real estate decelerating from the stepped-up pace of the last three years assumptions as a starting point, our REIT and core estimates reflect even as demand is expected to remain firm going forward. the same regional real estate fundamentals. The adjustments fine- Capital inflows should continue, reflecting healthy fundamentals tune these estimates. For example, 20% of underlying assets in and the underexposure of U.S. and EMEA institutional investors to regional REIT indices vary from the assets in that region’s core real the region. estate composite; the slightly higher growth potential for this 20% The developing services sector is expected to grow faster than adds an incremental return to our REIT assumptions. Our 2018 overall GDP, particularly in the China and India markets. Together methodology differs from previous years primarily by the inclusion with increasing urbanization, this should continue to drive of average industry leverage, which can also add to REIT returns commercial real estate growth. The changing complexion of the and have a differential impact across regions. (Exhibit 15). demographics in the region, both from a growing number of millennials and an aging population, should increasingly impact traditional real estate utilization assumptions. Diminished overcapacity, primarily in China, is expected to drive more efficient use of both office and logistic assets.

REIT return estimates assume convergence to the value of the underlying real assets and incorporate leverage EXHIBIT 15: REIT RETURN ASSUMPTIONS AND BUILDING BLOCKS (LEVERED, LOCAL CURRENCY, %)

REITS U.S. Europe* Asia Pacific Global Core real asset assumption 5.25 5.50 5.50 5.33 Net leverage benefit 0.40 1.25 1.10 0.66 NAV amortization 0.35 0.25 0.40 0.35 REIT vs. real asset composite difference 0.25 0.00 0.00 0.17 Total return 6.25 7.00 7.00 6.50 Source: J.P. Morgan Asset Management; estimates as of September 30, 2017. * Includes European core prime and non-prime asset assumptions.

J.P. MORGAN ASSET MANAGEMENT 75 ALTERNATIVE STRATEGY ASSUMPTIONS

INFRASTRUCTURE EQUITY

Our long-term outlook is for infrastructure equity returns of The asset class outlook reflects the same underlying intermediate 6.25%. While unchanged from 2017, our 2018 assumption reflects themes as our 2017 Long-Term Capital Market Assumptions—an a slight but offsetting shift in the building blocks of return: We expectation that regulators will become less reluctant to authorize anticipate a decline in valuation impact as strong demand bids up capital expenditures at higher rates of return as the necessity of a project pricing, and expect an equivalent increase in average yield replacement cycle, particularly in the U.S., becomes more acute. from investors requiring a higher illiquidity premium for these An accumulating capital spending deficit is hard to deny, though its long-term investments (Exhibit 16). expected size depends on the source of the estimate (Exhibits 17A and 17B). At the same time, the ongoing surge in demand No change in infrastructure equity returns, but a shift continues for stable high cash flow assets attractive to multiple in components investor types—from sovereign wealth funds to pensions and EXHIBIT 16: OECD INFRASTRUCTURE EQUITY—RETURN ASSUMPTIONS AND sophisticated individual investors. Nevertheless, higher returns are BUILDING BLOCKS (USD, %) likely to be required to galvanize funding for this plethora of infrastructure projects as they come on line. There are, however, 2018 2017 challenges to catalyzing momentum in the short term, as well as Valuation impact 1.00 1.25 elements of uncertainty in the longer term. Average yield 3.50 3.25 OECD/developed inflation 1.75 1.75 In the short term, the outlook for regulatory cycle improvement Total return 6.25 6.25 and a meaningful acceleration of infrastructure spending appears

Source: J.P. Morgan Asset Management; estimates as of September 30, 2016 and to be on hold. Consistent with the breakout of populist politics, the September 30, 2017. tone of the regulatory environment has gotten tougher on the margin and allowed returns on equity (ROEs) have been stable to down, though still consistent with a lower weighted average cost Core outlook of capital. The attractiveness of infrastructure as an investment lies in its At the far end of our 10- to 15-year LTCMA time frame, the view potential to provide stable and relatively high core returns is clouded by the difficulty of anticipating the various ways in augmented by the inflation pass-through accorded selected which technological innovation, populist sentiment and the sectors. The overall economic growth and inflation outlook— inclusion of public-private partnerships (PPPs) may impact modest and steady—should continue to be the primary infrastructure economics. determinant of core infrastructure returns and their stability over much of our forecast period.

Delayed investment in aging infrastructure portends a very significant capital spending gap EXHIBIT 17A: U.S. STATE AND LOCAL INVESTMENT IN CAPITAL PROJECTS AS EXHIBIT 17B: ESTIMATED U.S. INFRASTRUCTURE INVESTMENT GAP FOR A PERCENTAGE OF GDP (%) 2016–25 PERIOD (CUMULATIVE GAPS, BILLIONS OF CONSTANT 2015 DOLLARS)

Investment as % of GDP Average Ports and inland waterways $ 15 % 2.75 Aviation $42

2.50 Water and wastewater $105

2.25 Electricity $177

2.00 Surface transportation $1,101

1.75

1.50

’06 ’08 ’10 ’12 ’14 ’16 2016 ESTIMATE

Source: U.S. Bureau of Economic Analysis; data as of June 30, 2017. Source: American Society of Civil Engineers, 2016 Failure to Act analysis.

76 LONG-TERM CAPITAL MARKET ASSUMPTIONS A STABLE TO IMPROVING ALPHA OUTLOOK OFFSETS THE BETA DRAG

A surge in activity based on the injection of private capital into the Renewable energy is expanding its role as a major source for infrastructure demand/supply mix appears unlikely for the time electricity generation being, at least in the U.S. The sale of, and contractual agreements EXHIBIT 18: U.S. NET ELECTRICITY GENERATION BY FUEL (BILLION KILOWATT HOURS) related to, taxpayer assets are not always well received by the Coal Petroleum Natural gas Nuclear Renewable stakeholders, such as affected employees, labor unions and the sources BkWh 2016 public utilizing the asset, and fair sharing of the risks and rewards 2,500 is often seen through a political lens. In short, while the potential HISTORY PROJECTIONS for infrastructure asset growth is significant, it is difficult to insert 2,000 private equity ownership and terms into the existing structure of regulation even before the emotional obstacles of populism and 1,500 partisan politics come into play. 1,000 Regulated power 500 The most stable part of the infrastructure complex—electric 0 distribution—is undergoing a longer-term adjustment as the ’80 ’90 ’00 ’10 ’20 ’30 ’4 0 industry begins to feel the impact of technological and societal Source: U.S. Energy Information Administration Annual Energy Outlook 2017, change. Consider the small yet growing impact of local sources of J.P. Morgan Asset Management. alternative power generation feeding into utilities’ power grids. Utilities are under pressure to make regulator-approved capital improvements and expenditures in order to bring supply from distributed generation and renewables, especially solar, onto the INFRASTRUCTURE DEBT grid. Yet their ability to recoup these investments—through, for Our long-term return assumption for infrastructure debt is example, “smart metering”—may be necessary to allow the unchanged from last year’s projection. industry to transition from the current system. On the positive side, renewable energy is projected to expand its role as a leading We view infrastructure debt as essentially an A credit (based on source for electricity generation (Exhibit 18). Upgrade of the long-term credit loss statistics from rating agency data) with a BBB distribution grid is necessary longer term to accommodate the yield. The spread above the extrapolated credit rating represents a new sources of power, general obsolescence and growth. This premium demanded by investors for the relative illiquidity of should result in an uptick in regulator-allowed returns on those infrastructure debt with more than seven years to maturity. The new assets. illiquidity spread emanates from the original sourcing of debt as project finance loans underwritten by banks and generally kept on Transportation their books. The increasing secondary market liquidity and secondary market issuance of these loans may eventually reduce This sector is directly impacted by the consumer outlook, as the the required illiquidity premium, but until there is broader improved cyclical economic conditions and lower oil prices are investment experience with these strategies, the required illiquidity passing through to a better pace of demand for toll roads and premium should remain somewhat elevated at approximately consumer-related transport generally. Seaports are experiencing a 100bps (Exhibit 19). slower demand outlook as China’s growth trajectory slows and global trade has decelerated. Infrastructure debt—an A credit with a BBB yield Contracted power EXHIBIT 19: GLOBAL INFRASTRUCTURE DEBT—RETURN ASSUMPTIONS AND BUILDING BLOCKS (USD, %) In an industry where the term of contracts is a key value indicator, 2018 2017 the recent lengthening of contract terms is a distinct positive, as the percentage of the time an asset is owned outside of the A rated credit total return assumption 3.25 3.25 contract term exposes the operator to price risk. The renewables Required illiquidity premium 1.00 1.00 boom, however, is lowering the average cost of power generation Total return 4.25 4.25 and consequently driving down revenue growth. Source: J.P. Morgan Asset Management; estimates as of September 30, 2016 and September 30, 2017.

J.P. MORGAN ASSET MANAGEMENT 77 ALTERNATIVE STRATEGY ASSUMPTIONS

COMMODITIES

Our 2018 long-term compound annual commodity return BUILDING BLOCKS OF assumption of 3.75% implies a positive real return of 150 basis COMMODITY RETURNS* points over our U.S inflation outlook (2.25%). Projected returns We build our assumptions for commodity returns on the (real and nominal) are unchanged from last year’s assumptions. In long-term record of the Bloomberg Commodity Total Return fact, all of the building blocks we use in constructing our Index (a collateralized, investible index). We start with our commodity assumptions, starting with the U.S. inflation outlook, LTCMA for U.S. inflation and adjust for: are unchanged from 2017 (see “Building blocks of commodity (1) the differential between the 25-year collateralized returns”). commodity index return and inflation in the U.S. (which we estimate netted to 0.00 for the 1991-2015 period, Current state of the commodity cycle despite interim differentials)

Most current indicators continue to point to an upturn in the broad (2) where we are in the current commodity cycle (pricing theories based on the economics of non-renewable commodity cycle following its turnaround from the bottoming of the resources in finite supply are not embedded in our commodity supercycle over a year ago. However, the energy sector, estimates) accounting for roughly 30% of the Bloomberg Commodity Total Return Index, presents an exception. While an uptick in global (3) a scaling effect to account for the absolute increment in commodity usage of key marginal consumers (namely, economic activity has stabilized energy demand and the OPEC emerging Asian and frontier economies) production agreement of November 2016 is helping to constrain supply, the innovative and low cost U.S. fracking industry has seen (4) the inverse relationship between commodity returns and the U.S. trade-weighted dollar these developments as an invitation to loosen production discipline. As a consequence, the energy segment has now taken on late-cycle (5) the potential contribution from roll yields (which we characteristics in terms of supply outpacing demand, falling prices, found for the 1994-2015 period to be inconsistent and easy access to expansion capital and energy operators’ seeming statistically negligible; we expect a zero contribution from this source during the 10- to 15-year time frame of expectation of permanent demand absorption. our assumptions.) We see the fracking industry’s response as delaying and muting the energy sector’s cyclical upturn, but only in the initial years of * For further details on our methodology, please see J.P. Morgan Asset Management, 2017 Long-Term Capital Market Assumptions, pp. 71-73. our forecast period. Arguably, fracking price curves (Exhibits 20A and 20B) enable the industry to produce more oil and gas at a declining cost per barrel, but the incentive to add to current

Shale drilling innovation enables more oil production per rig at a decreasing cost per barrel EXHIBIT 20A: U.S. NEW-WELL OIL PRODUCTION PER RIG EXHIBIT 20B: OIL PRODUCTION—KEY U.S. SHALE PRODUCERS VS. AVERAGE SHALE WTI BREAKEVENS bbl/day Average breakeven, U.S. dollar/bbl (LHS) U.S. production, Mbbl/day (RHS) 800 USD/bbl Mbbl/day 80 14 700 70 600 12 500 60

400 50 10 300 40 200 8 30 100

0 20 6 ’07 ’08 ’09 ’10 ’11 ’12 ’13 ’14 ’15 ’16 ’17 ’13 ’14 ’15 ’16 ’17E ’18E ’19E ’20E ’21E ’22E ’23E ’24E

Source: U.S. Energy Information Administration; data as of June 2017. Source: Company reports, U.S. Department of Energy, U.S. Energy Information Administration, J.P. Morgan Asset Management; data as of July 2017.

78 LONG-TERM CAPITAL MARKET ASSUMPTIONS A STABLE TO IMPROVING ALPHA OUTLOOK OFFSETS THE BETA DRAG excess supply assumes sufficient demand to absorb these Refining the estimation process increases at an attractive price. Current demand/supply conditions seem inconsistent with higher oil prices on a cyclical basis unless The Bloomberg Commodity Total Return Index is basically the economic cycle grows past the excesses of current supply. In unchanged from its level at the time of last year’s LTCMA our view, that growth will materialize, absorbing current excess publication—masking widely divergent sectoral returns. In supply and allowing the energy cycle to resume its upturn. examining the component pieces of the index over the past year, we find both rapidly changing cyclical and long-term drivers On balance, considering Chinese basic metals capacity shutdowns, behind the pattern of recent returns. Analyzing the reasons behind OPEC supply constraint and U.S. fracking, among other factors, the short-term cyclical volatility presents additional insight into the outlook for the broad commodities market continues to be one of estimation methodology and points to potential refinements to the modest supply rationalization. The supply rationalization process, estimation procedure. as indicated by our Commodity Event Index (see “The Commodity Event Index—Capturing producers’ supply constraint sentiment”), For example, we have made slight adjustments to our Commodity while not quite as robust as it appeared last year, is still essentially Event Index, which seeks to capture the underlying drivers of operative relative to the demand outlook. As such, we make the supply rationalization as a key element in projecting long-term same incremental adjustment (+.25bps) as we did last year to our cycle dynamics. Since the technological innovation within the U.S. commodity return assumption for where we are in the current shale industry is having a heightened impact on global oil commodity cycle (Exhibit 21). production and the overall commodity change cycle, a reweighting of the component pieces of the index, including an increase in the weighting of the explicit energy rig count factor, is warranted. In effect, with the reweighting of the index, the new commodity cycle appears slightly less robust in turning the corner on past excesses.

Our commodity return assumption is 1.50% in excess of U.S. inflation EXHIBIT 21: COMMODITIES—RETURN ASSUMPTIONS AND BUILDING BLOCKS (USD, %)

2018 2017 U.S. inflation assumption 2.25 2.25 Adjustment for historical 25-year investible index return above inflation 0.00 0.00 Position in current cycle (premium/discount) 0.25 0.25 Scaling function adjustment for emerging and developing Asian economies 0.25 0.25 USD decline impact (projected incremental annual decline vs. historical base period) 1.00* 1.00 Impact of roll yield 0.00 0.00 Total return 3.75 3.75

Source: J.P. Morgan Asset Management; estimates as of September 30, 2016 and September 30, 2017. * Rounded up from 0.90.

J.P. MORGAN ASSET MANAGEMENT 79 ALTERNATIVE STRATEGY ASSUMPTIONS

THE COMMODITY EVENT INDEX—CAPTURING PRODUCERS’ SUPPLY CONSTRAINT SENTIMENT THE COMMODITY EVENT INDEX

6.0

5.0 -3 SD

+2 SD 4.0 2018 index value +1 SD 3.0 Mean -1 SD 2.0 -2 SD 1.0 +3 SD

0.0 ’85 ’90 ’95 ’00 ’05 ’10 ’15

COMMODITY EVENT INDEX COMPONENTS The Commodity Event Index is designed to capture producer sentiment around the loosening/tightening of production constraints within commodity markets. Higher index values indicate a more constrained environment, supportive of increasing commodity prices. The event index utilizes a component weight scheme in which four components have 11.1% weightings, while three components that we deem more important receive an 18.5% weighting, as indicated below. Components were added as available (inclusion date in parentheses) for our universe of energy and materials companies, including:

Observed change to Impact on Index component Component weight % index component index value Credit rating (1985) 11.1 lower higher Age of capital stock (1985) 11.1 older higher Financial leverage (1985) 11.1 higher higher Volume of bankruptcies, takeovers, debt-for-equity swaps (2004) 11.1 higher higher Capital expenditure to sales (1985) 18.5 higher higher Oil rig count (1991) 18.5 higher lower CEO turnover (2007) 18.5 higher higher

Source: Baker Hughes, Bloomberg, U.S. Bureau of Economic Analysis, J.P. Morgan Asset Management; Commodity Event Index estimation as of May 31, 2017.

Gold

The return for gold is driven by many of the same factors as accumulation of gold for investment purposes. Within the last few general commodity returns but primarily by U.S. inflation, the years, central banks have ceased liquidating their gold reserves direction of the trade-weighted U.S. dollar and a scaling factor and have started accumulating once again. We project a 25bps that reflects the increasingly important developing economy gold return premium to broad commodities (equivalent to last impact on gold consumption. Consumption per capita can be year’s), implying a 4.00% return for gold. expected to fall in China and India. However, since the two highest per capita gold consumers (with roughly twice the per capita consumption of developed economies) are also the two fastest- growing economies, we expect the net effect to be an increase in the absolute demand for gold. Another small increment to demand is assumed from an erratic but still long-term

80 LONG-TERM CAPITAL MARKET ASSUMPTIONS CURRENCY EXCHANGE RATE ASSUMPTIONS

The tide has turned for the U.S. dollar Michael Feser, CFA, Portfolio Manager, Multi-Asset Solutions Jonathon Griggs, Head of Applied Research, Global Fixed Income, Currency & Commodities

IN BRIEF • Overvaluation of the U.S. dollar, which had reflected divergences in the cyclical positions between the U.S. and other economies, has begun to unwind amid low inflation and a Federal Reserve moving in slow motion. Unwinding USD strength historically takes around seven years, and we see continued USD weakness over our forecast horizon, even after the retracement from last year’s levels.

• While the euro has moved off its cyclical lows, it remains well below fair value. We expect that eurozone economic data—strengthening growth, current account surplus and inflation below the U.S. level—will boost the euro over our forecast horizon.

• Uncertainty surrounding the outcome of UK negotiations to exit the European Union (EU) suggests that forecasts for sterling should reflect a modest increase in Brexit-related risk premia as well as an increase in the downside risks to the UK economy over the medium term. We have slightly lowered our long-term fair value assumption for sterling.

J.P. MORGAN ASSET MANAGEMENT 81 CURRENCY EXCHANGE RATE ASSUMPTIONS

FAIR VALUE AND SECULAR INFLATION AND GROWTH As a consequence, most of our fair value exchange rate TRENDS assumptions are little changed from last year, even if considerable adjustments may be required in future years, Compared with last year’s Long-Term Capital Market Assumptions depending on the actual economic trajectory of the UK (LTCMAs), in this year’s edition the longer-term secular inflation and China. and growth trends, impacting the future fair value of currency exchange rates, have generally remained well entrenched, both on an absolute level for each countryas well as on a relative basis LONG-TERM CURRENCY EXCHANGE RATE between countries. ASSUMPTIONS

As in prior years, we have determined today’s fair value exchange In last year’s edition, we noted that the significant rise of the U.S. rates for G10 currencies through a relative purchasing power parity dollar was not supported by a change in its long-term fair value, (PPP) approach, based on the long-term average of each currency’s but was rather a result of divergences in the cyclical positions real exchange rate. To calculate the fair value for emerging market between the U.S. and other developed and emerging market (EM) currency exchange rates, we take an absolute PPP-based economies. We also expected that those divergences were close to approach that uses “actual individual consumption estimates,” their peak and that the Federal Reserve (Fed) would move very based on the analysis conducted by the World Bank and the slowly and carefully when normalizing policy rates. Organization for Economic Co-operation and Development for their international price comparison program. Over the last 12 months, those divergences have declined modestly as the synchronization of global growth dynamics has To arrive at a given exchange rate projection over the assumption rapidly gathered pace, developed market growth has come in well horizon, we have adjusted today’s fair value exchange rate using above potential, and labor market slack has been diminishing our assumptions for inflation and growth. For G10 currencies, we quickly. At the end of the third quarter, both global indicators and reflect the expected change in a country’s terms of trade over the financial conditions remained supportive for labor markets to assumptions horizon by adjusting today’s fair value for the tighten further. inflation rate differential between the countries. For emerging markets, we also make an additional adjustment for the expected The Fed’s hiking cycle is clearly more advanced than other central differential in the GDP per capita growth. banks’, but global central bank language, especially from the European Central Bank (ECB), has taken on a more hawkish tone. In our growth assumptions, we project that deteriorating demographics, smaller total factor productivity improvements and Somewhat surprisingly, core CPI inflation has remained low and is lower levels of human capital development will continue to hold not responding to labor market tightening in the manner back developed market (DM) economies. Emerging markets in traditional central bank models would predict. With the notable aggregate will continue to grow faster than their DM counterparts, exception of the UK, G4 central bankers have recently lowered given larger increases in the size and quality of their labor forces, their inflation forecasts, voicing concern that inflation will struggle although with a wide dispersion. Russia, Taiwan and China, in to reach their targets. We therefore continue to expect the path of particular, are likely to experience labor force shrinkage in the normalization in all economies to be both slow and shallow. coming years. As a result, and a little earlier than we had expected, this dynamic We continue to assume that over the LTCMA horizon developed has allowed the overvaluation of the U.S. dollar to begin to market central banks will generally come close to achieving their unwind—a process that historically tends to last, on average, inflation targets, with shorter-term fluctuations occurring within a around seven years (Exhibit 1). Stronger growth in EM economies, narrow band. Even for emerging market economies, coupled with a Fed in slow motion, is also bullish for EM we expect a relatively stable inflation environment, albeit at levels currencies. With domestic reforms unfolding gradually in Latin somewhat above their respective central bank targets. America, most of the emerging market growth surprises are coming from Asia. The primarily political threats to our assumptions for the eurozone and the U.S. appear to have receded since last year’s edition. Sterling, impaired by political uncertainty and fears over a hard However, Brexit has clouded the UK’s prospects, and China’s ability Brexit, was the exception to this trend—it weakened in the to transition smoothly from an investment-led to a more balanced aftermath of June 8 parliamentary elections. growth model is far from assured.

82 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE TIDE HAS TURNED FOR THE U.S. DOLLAR

Our assumptions continue to point toward future U.S. dollar weakness, even after the retracement from last year’s levels EXHIBIT 1: ASSUMPTIONS FOR SELECTED CURRENCY EXCHANGE RATES—NEXT 10-15 YEARS (According to market convention, CURRENCY A/CURRENCY B means one unit of CURRENCY A is worth the stated number of units of CURRENCY B. EUR/USD = 1.30 means EUR 1.00 is worth USD 1.30.) Current levels 2018 2017 Per annum FX rate FX rate Currency September 30, 2017 % appreciation vs. USD assumptions assumptions Euro EUR/USD 1.18 +1.00 1.34 1.31 Japanese yen USD/JPY 112.51 +1.50 93 89 Swiss franc USD/CHF 0.97 +0.75 0.88 0.88 British pound GBP/USD 1.34 +0.75 1.47 1.52 Canadian dollar USD/CAD 1.25 +0.75 1.14 1.12 Australian dollar AUD/USD 0.78 -0.75 0.71 0.74 Chinese renminbi USD/CNY 6.65 +1.00 5.87 6.07 Brazilian real USD/BRL 3.16 -1.00 3.59 3.94 Mexican peso USD/MXN 18.26 +1.25 15.63 16.65

Source: Bloomberg, J.P. Morgan Asset Management; estimates as of September 29, 2017. *For consistency and ease of conversion, we have assumed that the forecast horizon for the per annum change in percentage terms is 12.5 years.

Euro Japanese yen

Using an expanded array of monetary tools, the ECB has achieved a Japanese real economic growth remains at the lower end of the significant reduction in funding costs, materially improving the fiscal developed market range as a challenging demographic outlook position of the EU region’s governments and aiding the cyclical dominates recent improvements in productivity. In the past year, economic recovery. As a result, and contrary to the expectations of the Bank of Japan (BoJ) broke new policy ground by shifting many market participants, the euro area is following the U.S. toward a more targeted monetary policy, using yield curve control recovery road map, with about a three- or four-year lag. Delayed by as its central tool to maintain an extremely easy monetary regime. a second recession after the Great Recession, the euro area’s economic growth cycle is gaining traction and is increasingly self- Record overall employment implies an economy close to full reinforcing, including in much of the periphery. Pro-EU election capacity even with an apparent low level of GDP growth. results, particularly in France, have reduced political uncertainty to We therefore expect that Japanese inflation rates will, on average, a considerable extent and have provided political capital to focus on fall well short of the BoJ’s 2% target. Financial repression will institution-building and domestic reforms. therefore still be necessary to erode Japan’s high level of sovereign debt throughout our forecast horizon. Not surprisingly, ECB rhetoric now signals a less dovish approach, with an initial focus on ending the use of the central bank’s As a result, the yen has unwound some of its prior strength and at politically most controversial policy tool, quantitative easing (QE). its current level remains significantly below our estimate for its Actual interest rate normalization appears some ways off, as long-term fair value. Given the need for an easy monetary policy inflation levels remain subdued and well below the ECB’s target. environment throughout our forecast horizon, we expect the yen to trade closer to, but below its fair value at USD/JPY 93. As a result, the euro has recovered further from its cyclical lows, but it is still well below fair value at EUR/USD 1.18 as of the end of Swiss franc September. Over the assumption horizon, however, we expect that The Swiss franc has continued to trade in a narrow range just the eurozone’s current account surplus, coupled with an inflation above our long-term fair value estimate. Over the LTCMA horizon, rate below the U.S. level, will result in a 1.00% annual we expect the Swiss franc to continue to benefit from a relatively appreciation of the euro to the equivalent of a EUR/USD 1.34 more benign inflation outlook than the U.S., suggesting a rise at exchange rate. a long-term annualized rate of 0.75% against the dollar to USD/ CHF 0.88.

J.P. MORGAN ASSET MANAGEMENT 83 CURRENCY EXCHANGE RATE ASSUMPTIONS

Sterling Commodity currencies Sterling is the most difficult of the G10 currencies to assess, given As commodity prices and inflation have bottomed out, the range of potential outcomes for Brexit negotiations. Although the Canadian dollar has recovered from some of its prior we have not changed our long-term growth and inflation underperformance relative to the U.S. dollar. Nevertheless, there projections for the UK from last year, we note that these represent is still a need for domestic rebalancing in an economy that a central case scenario and that the risk distribution on both sides accumulated substantial amounts of household credit. Preparing a has continued to widen. This suggests that investors should soft landing for a highly overheated housing market is another continue to attach a larger than normal risk premium on sterling unresolved challenge. The Canadian dollar may therefore have to whatever the actual data for relative inflation or growth over the trade at a discount relative to fair value for some time longer. In medium term. fact, the loonie continues to trade below our long-term fair value assumption of USD/CAD 1.14, and we therefore expect it to In recent years, our estimates for the prevailing PPP-based fair appreciate by 0.75% per annum. value for GBP/USD have been close to 1.60. Because our macro assumptions imply no underlying inflation differential with the U.S. Despite its exposure to the Chinese economic slowdown, elevated over the next 10 to 15 years, our forward-looking PPP assumption house prices and typical sensitivity to a change in the U.S. rate therefore remains close to the 1.60 level as well. Last year we cycle, the Australian dollar is trading at the upper end of its narrow incorporated a small Brexit-related negative risk premium for range, close to its 2003 levels and somewhat above our long-term sterling into our calculations and acknowledged the risk that a assumption of AUD/USD 0.71. A near-term undershoot to the weaker shorter-term economic outlook would depress sterling for downside, similar to that already experienced by the Canadian a more prolonged period. dollar, has become less likely, but still presents a risk, given the Developments since the UK election in June 2017 suggest that the domestic economic imbalances. Brexit-related risk premium on sterling should be increased a little Brazil faces a new round of political turmoil as its focus shifts from further, and that the downside risks to the economy in the dealing with the economic fallout from the decline in commodity medium term have also increased. We have therefore lowered our prices over 2014-15. We have moderately improved our expected long-term fair value assumption modestly to 1.47. From current inflation trajectory, but currency markets continue to discount levels, this implies a smaller 0.75% rise in sterling over our substantially more progress on the inflation front than we forecast horizon, compared with the 1.25% rise incorporated in consider sustainable over the LTCMA forecast horizon. We last year’s LTCMAs. therefore expect a somewhat weaker Brazilian real, falling to a rate of USD/BRL 3.59 vs. the current spot level of 3.16.

BUILDING BLOCKS–CURRENCY EXCHANGE RATES The annualized compound rate of change expresses the difference between two currencies’ current exchange rate and our estimate of their fair value exchange rate at the end of our assumptions horizon—for consistency we use 12½ years. A DEVELOPED MARKETS B EMERGING MARKETS • Starting fair value exchange rate based on the theory of • Starting fair value exchange rate based on the theory of purchasing power parity (PPP) purchasing power parity (PPP) + Expected future inflation rate differential between domestic + Expected future inflation rate differentials and GDP per economies capita growth differentials* between domestic economies + Review qualitatively and adjust currencies selectively to + Review qualitatively and adjust currencies selectively to ensure internal consistency and incorporate secular factors ensure internal consistency and incorporate secular factors and trends other than relative inflation that would otherwise and trends other than relative inflation that would otherwise not be captured not be captured + The prevailing spot exchange rate level on September + The prevailing spot exchange rate level on September 29, 2017 29, 2017

* Academic studies suggest real equilibrium exchange rates in emerging economies are enhanced via the convergence process of higher productivity and trend growth rates. This can be proxied by GDP per capita. See Choudri and Khan (2004), “Real Exchange Rates in Developing Countries: Are Balassa-Samuelson Effects Present?”, IMF Working Papers. Kravis and Lipsey (1983), “Toward an Explanation of National Price Levels,” Princeton Studies in International Finance.

84 LONG-TERM CAPITAL MARKET ASSUMPTIONS THE TIDE HAS TURNED FOR THE U.S. DOLLAR

In Exhibit 2, we present selected exchange rate histories relative to our 2018 LTCMAs.

EXHIBIT 2: SELECTED EXCHANGE RATE HISTORIES RELATIVE TO 2018 LTCMAs

EUR 2018 LTCMA JPY 2018 LTCMA CHF 2018 LTCMA

1.6 150 2.1

1.9

1.4 130 1.7

1.5 1.2 110 1.3

1.1 1.0 90

0.9

0.8 70 0.7 ’93 ’01 ’09 ’17 ’93 ’01 ’09 ’17 ’93 ’01 ’09 ’17

GBP 2018 LTCMA CAD 2018 LTCMA AUD 2018 LTCMA

2.1 1.7 1.2

2.0 1.6

1.9 1.5 1.0 1.8 1.4

1.7 1.3 0.8 1.6 1.2

1.5 1.1 0.6 1.4 1.0

1.3 0.9

1.2 0.8 0.4 ’93 ’01 ’09 ’17 ’93 ’01 ’09 ’17 ’93 ’01 ’09 ’17

BRL 2018 LTCMA MXN 2018 LTCMA USD (J.P. Morgan U.S. CPI-based real broad eective exchange rate) 2018 LTCMA 25 130

4.0 20 120

3.0 15 110

2.0 10 100

1.0 5 90

0.0 0 80 ’93 ’01 ’09 ’17 ’93 ’01 ’09 ’17 ’93 ’01 ’09 ’17

Source: Bloomberg, J.P. Morgan Asset Management; data as of September 30, 2017. The annualized compound rate of change expresses the difference between the two currencies’ current exchange rate and our estimate of their fair value exchange rate at the end of our assumptions horizon-for consistency we use 12 ½ years.

J.P. MORGAN ASSET MANAGEMENT 85 VOLATILITY ASSUMPTIONS

Volatility: Cyclically lower, structurally unchanged Grace Koo, Ph.D., Quantitative Analyst and Portfolio Manager, Multi-Asset Solutions Nandini Srivastava, Quantitative Analyst, Multi-Asset Solutions Livia Wu, Quantitative Analyst, Multi-Asset Solutions

IN BRIEF • Our long-term volatility assumptions are largely unchanged from 2017, despite the low level of asset volatility across markets generally.

• In our view, the low levels of volatility across asset classes are consistent with the late stage of the business cycle and not the result of a structural change or thematic shift. Therefore, we do not adjust our volatility assumptions downward relative to last year’s estimates.

• Based on our long-term return and volatility assumptions, Sharpe ratios, in general, are projected to fall marginally for equities and credit while increasing for government bonds. Despite the decline, credit continues to rank well in expected risk-adjusted return terms, closely followed by equities. The risk-adjusted return for government bonds remains considerably below its historical average.

• This year we introduce two enhancements to our process for estimating volatility, designed to ensure that our assumptions process has the flexibility to incorporate our forward- looking expectations, whether those views are in line with or deviate from historical business cycle length and composition.

86 LONG-TERM CAPITAL MARKET ASSUMPTIONS VOLATILITY: CYCLICALLY LOWER, STRUCTURALLY UNCHANGED

LITTLE CHANGE IN RISK ASSUMPTIONS We also adjust our assumptions for emerging market debt (EMD) volatility upward vs. its historical trend. This reflects our Our 2018 Long-Term Capital Market Assumptions (LTCMA) volatility expectation that the credit rating migration of many emerging projections are generally unchanged from 2017 estimates, despite markets from below investment grade to above investment grade a currently low asset volatility environment. Clearly, volatility in witnessed over the past decade may partially reverse itself during financial markets has been running at historically low levels, our forecast period. particularly for risk assets and most notably for equities (Exhibit 1). This low volatility may prompt some to ask if structural changes Short-duration instruments have been abnormally stable in in asset volatility have taken place. As we will show, we find little recent years as central bank interventions created distortion in evidence to support this conjecture. In our view, currently low the market. As central bank stimulus is gradually removed and asset volatility levels are in line with the late phase of the business policy rates rise, the subdued volatility of short-duration cycle we are in—and we see little need to adjust our forward- instruments will likely revert to longer-term levels, above what looking estimates downward. recent history would suggest.

In contrast, we expect European high yield debt to be less volatile Asset volatility is running at historical lows—especially in equities going forward. The quality of the market has improved in recent EXHIBIT 1: ONE-YEAR ROLLING REALIZED VOLATILITY OF GLOBAL EQUITIES AND FIXED INCOME (%)* years, and fallen angels are expected to regain their investment U.S. large cap equities U.S. small cap equities EAFE equities grade status over our LTCMA forecast period. German equities U.S. ﬁxed income (RHS) % 60 7 Implications for risk-adjusted returns across asset classes 6

5 Combining our LTCMA risk and return assumptions, we compare 40 4 the expected return per unit of risk taken (as measured by the Sharpe ratio)1 among asset classes. Since we developed last year’s 3 20 LTCMAs, we have seen a strong rally in equities and risky assets 2 and a mild move up in rates. With these starting points, this year’s 1 return assumptions are generally lower for equities and credit and

0 0 slightly higher for government bonds. Given the negligible ’90 ’99 ’08 ’17 movement in volatility assumptions, Sharpe ratios, in general, are Source: Bloomberg Barclays, Deutsche Börse AG, FTSE Russell, MSCI, Standard & projected to fall marginally for equities and credit while increasing Poor’s, J.P. Morgan Asset Management; data as of July 31, 2017. for government bonds. Despite the decline, credit continues to *U.S. large cap equities based on S&P 500 returns; U.S. small cap equities based on Russell 2000 returns; EAFE equities based on MSCI EAFE returns; German equities based rank well in expected risk-adjusted return terms, closely followed on DAX returns; U.S. fixed income based on Bloomberg Barclays U.S. Aggregate returns. by equities. The risk-adjusted return for government bonds remains considerably below its historical average. Where history may not be the best guide

For most of the assets in our LTCMA opportunity set, including WHY RECENT LOW ASSET VOLATILITY DOES NOT equities, we expect long-term volatilities to be in line with history. CHANGE OUR LONG-RUN EXPECTATIONS However, there are several fixed income sectors for which our risk A historical examination of asset volatility suggests to us that recently estimates are different (mostly higher) than what the past decade low volatility levels are not indicative of a structural change or or so of history may suggest. thematic shift requiring adjustments to our long-term assumptions. The credit quality of U.S. and European investment grade (IG) Volatility in the macroeconomic environment (as measured by corporate bond issuers has been gradually declining over the past changes in real GDP) has been declining for several decades, 10 years, with AAA rated companies becoming a rarity. In the U.S. smoothed by improvements in monetary policy, broader and elsewhere, the majority of IG bonds had an A rating in the availability of consumer credit and the effects of globalization. early 2000s, whereas the majority now have a BBB rating. We adjust our forward-looking volatility estimates upward relative to historical levels for these fixed income sectors. 1 The Sharpe ratio is a measure used to examine the performance of an investment by adjusting for its risk. The Sharpe ratio is the return in excess of the risk-free rate per unit of risk.

J.P. MORGAN ASSET MANAGEMENT 87 VOLATILITY ASSUMPTIONS

However, as Exhibit 2 suggests, the likelihood of asset bubbles measured by the CBOE VVIX Index3—the option-implied volatility of and financial stress is not necessarily diminished by a smoother VIX) has remained pronounced and dynamic, responding at the macro environment. Despite the downtrend in macro volatility first signs of stress (Exhibit 4). since the 1980s, market volatility has stayed close to its long- run average. In short, the recent low asset volatility environment may have edged our long-term forecasts a touch lower for some asset classes, but, on balance, the broad outlook remains the same as Asset volatility remains near its long-run historical average, despite last year’s. a downtrend in macro volatility EXHIBIT 2: MACRO AND MARKET VOLATILITY, THREE-YEAR ROLLING QUARTERLY CHANGES (%) Realized volatility in expansions is markedly lower than during Macro volatility S&P volatility Long-run average contractions (based on changes in real GDP) S&P volatility % EXHIBIT 3: S&P 500 12-MONTH REALIZED VOLATILITY (%) 30 Expansions Contractions 25 Expansions average Contractions average % 20 60

10 40

20 0 ’52 ’68 ’84 ’00 ’16

Source: Bloomberg, U.S. Bureau of Economic Analysis, Standard & Poor’s, J.P. Morgan Asset Management; data as of July 31, 2017. 0 ’27 ’37 ’47 ’57 ’67 ’77 ’87 ’97 ’07 ’17

Source: Bloomberg, National Bureau of Economic Research, Standard & Poor’s, In fact, benign macroeconomic volatility could create risks of J.P. Morgan Asset Management; data as of July 31, 2017. overvaluation in asset classes, an environment that could lend itself to creation of asset price bubbles and potentially generate greater volatility, offsetting concurrent low macro volatility. What’s more, Asset volatility may be low, but the volatility of volatility is not the extraordinary measures taken by global central banks may be EXHIBIT 4: AVERAGE 30-DAY SPX VOLATILITY INDEX (VIX) AND contributing to a more gradual and smooth recovery, but any NEAR-TERM EXPECTED VOLATILITY OF VIX (VVIX) (%) misstep in their unwinding could result in greater asset volatility. VVIX (LHS) VIX (RHS) % In our view, the recent reduction in asset volatility is consistent with 120 70 the current economic environment, absent any significant and persistent signs of stress or leverage that could create risks of contagion. Ninety years of history shows that volatility is typically 6 100 50 percentage points lower during expansions—averaging 15%—than during contractions (Exhibit 3).

In addition, we see little evidence that the current low asset 80 30 volatility regime (as measured by the CBOE Volatility Index [VIX])2 has unduly dampened the response of the asset markets to 60 10 events. That is, despite the fact that the VIX has been running at a 0 10 20 30 historically low level, the expected volatility of volatility (as Source: Bloomberg, Chicago Board Options Exchange (CBOE), J.P. Morgan Asset Management; data as of July 31, 2017.

2 The CBOE Volatility Index (VIX) is a measure of forward-looking near-term (30- day) market volatility and investor sentiment. Values greater than 30 are generally associated with a high level of volatility and investor fear; values less than 20 3 The CBOE VVIX Index is an indicator of the expected volatility of the 30-day forward generally correspond to a less stressful market environment. price of the VIX.

88 LONG-TERM CAPITAL MARKET ASSUMPTIONS VOLATILITY: CYCLICALLY LOWER, STRUCTURALLY UNCHANGED

WHAT IF THE LOW ASSET VOLATILITY ENVIRONMENT Historical asset volatilities are expected to decline as periods of PERSISTS? stress roll-off, replaced by periods of calm EXHIBIT 5: DECLINE IN ASSET VOLATILITIES ASSUMING A CONTINUED PROLONGED ECONOMIC EXPANSION (VOLATILITY OF ROLLING 10-YEAR Most forecasters would agree that the probability of a major MONTHLY RETURNS, %)* economy starting to contract is low over the next 12 to 24 months, Developed world equity U.S. large cap EAFE equity Emerging markets equity suggesting a further lengthening of the current expansion. We have % traditionally relied on the most recent 10 years of return data to 25 anchor our forward-looking risk expectations. We chose 10 years as Retains 2/3rds of stress periods our historical window because that period is recent enough for 20 relevancy and data availability and long enough to include at least 15 one whole business cycle. However, given the current extended recovery, it is possible that a 10-year window will soon be too short 10 to cover an entire cycle. This means that historical volatility may begin to understate future volatility as periods of contraction (and 5 high volatility, or “stress”) roll off and periods of recovery (low volatility, or “calm”) are added on.4 0 No Current 1 year forward 2 years 3 years stress forward forward periods Exhibit 5 attempts to simulate the impact of a prolonged expansion on historical volatility measures based on a fixed data Source: MSCI, Standard & Poor’s, J.P. Morgan Asset Management; data as of June 30, 2017. window. We start with a current 10-year period (mid-2007 to mid- * Developed world equity based on MSCI World returns; U.S. large cap equity based on S&P 2017) and roll forward (to mid-2008 to mid-2018 and then to mid- 500 returns; EAFE equity based on MSCI EAFE returns; emerging market equity based on 2009 to mid-2019). Early contractionary periods roll off, replaced MSCI EM returns. by our estimates, which are based on simulated scenarios reflecting a continued expansion. The result is an increasing To ensure our process has the flexibility to adapt to expectations proportion of calm periods in the sample set and a marked regardless of the historical business cycle length and composition decline in volatility, especially looking two to three years out. (periods of contraction vs. expansion),we incorporate an alternative way to anchor our forward-looking expectations in Expanding the data window helps mitigate this impact. Additional addition to using simple historical inputs. This modification analysis shows that, in general, including one incremental year of provides a more stable volatility forecast when the long-term risk data (i.e., moving from 10 to 11 years, 11 to 12 years, etc.) would expectation is for little change. It serves as an additional tool to lead to a more gradual decline in volatility of approximately 0.5 enhance the consistency of our process. percentage points with each additional year. This year we initiate the expansion of our data window, starting with one incremental year, and explicitly incorporate our forward- AN ALTERNATIVE WAY TO ANCHOR FORWARD- looking expectations of how frequently stress periods will occur. In LOOKING EXPECTATIONS essence, we identify the historical periods of higher relevance and overweight these data points to create an anchor that better The expected prolonged business cycle leaves us with two reflects our forward-looking view. (For more technical details, questions: please see “Volatility and correlation assumptions • Is 10 years long enough to capture full business cycle dynamics? methodology.”)

• Even if 10 years is long enough, does the proportion of high As discussed in prior sections, we expect the long-term risk volatility (stressed) to low volatility (calm) periods during that environment to persist because we see little evidence of structural historical interval reflect what we anticipate for the future? changes. Our simple historical (equally weighted) and our alternative (relevancy-weighted) anchors provide similar insights this year, as our forward-looking frequency of stress periods is in line with the 10-year history. If our expectation of a continued expansion materializes over the next one to two years, we would emphasize the stress/high volatility periods to ensure risk does 4 We define periods of stress (contraction) and calm (expansion) based on the not get underestimated, given the cyclical dynamics. National Bureau of Economic Research (NBER) business cycle reference dates.

J.P. MORGAN ASSET MANAGEMENT 89 VOLATILITY ASSUMPTIONS

VOLATILITY ASSUMPTIONS METHODOLOGY Long-term asset class volatilities and correlations tend to exhibit stability when measured over multiple cycles. As such, we use the following process in estimating risk assumptions for the main asset classes: 1. Start with monthly historical return data 4. Adjust for key themes and structural changes • In prior estimates, we used 10 years of historical data as the • Key themes and structural changes that are expected in the anchor. This year we increase the data window from 10 years forecast horizon, such as those highlighted in this article, are to 11 years. reflected in the long-term risk forecast accordingly. 2. Filter data outliers A few things to keep in mind: First, the standard deviation calculation is not subject to sequence risk. Thus, our assigned • Extreme data outliers are filtered to improve robustness. This is aggregate weighting of stress periods matters, but not the order done by winsorizing* historical raw data. of the data points or the continuity of the stress periods. Second, • For extreme data points exceeding the 99.5% threshold of a for consistency, the weights are applied to all the various currency normal distribution (a standard deviation of 2.58), we adjust the matrices we publish. The forward-looking periods and the treatment return data by setting it at the 99.5% threshold. of historical data are identical across regions and assets. Third, 3. Construct anchor matrix the volatility estimates capture the likely movement of the return around our central return forecasts. However, they do not • We leverage the historical experience to help anchor our incorporate distribution elements such as tail risk of the assets forward-looking expectations, focusing on: and other upper moments. It is particularly important for investors – simple historical return series (with each data point equally that hold assets known to have fat tails—such as high yield bonds, weighted) emerging market debt, convertible bonds, etc.—to account for risk aspects in addition to volatility. – historical return series with each data point weighted by “relevance” (the expected frequency of stress vs. calm For further discussion of our volatility and correlation process, periods)** please see: • Variance-covariance matrix is calculated using the filtered data “Creating more robust forward-looking risk statistics,” Daniel J. set. Scansaroli, and Michael Feser, J.P. Morgan Asset Management, 2015 Long-Term Capital Market Return Assumptions. – After filtering the data, we demean each data point by the average of the full sample. “Focusing on hedge fund volatility: Keeping alpha with the beta,” Daniel J. Scansaroli, Ph.D., J.P. Morgan Asset Management, – We multiply the weighted demeaned return time series November 2016. matrix to calculate the covariance matrix. – Volatility and correlation are extracted from the covariance matrix. The monthly volatility is then annualized by the industry standard square root of 12 factor.

* Winsorization applies a cap and a floor to extreme data values to remove the impact of potentially spurious outlier data on statistical results. ** We define stress periods based on NBER recession periods and assign them a long-run average probability of 15%. We apply these weights on a global basis. Although one may argue that in the future the U.S. may not be the sole dominant economy, over the historical sample period there was no instance in which an economic contraction in the U.S. left the global financial markets unaffected. Analysis based on a market definition of stress yields a similar outcome. Once we have adjusted for periods of stress/calm, we incorporate further forward-looking adjustments if needed.

90 LONG-TERM CAPITAL MARKET ASSUMPTIONS III Assumptions matrices

HOW TO USE THE NUMBERS Our assumptions can be used to: • Develop or review a strategic asset allocation • Understand the risk and return trade-offs across and within asset classes and regions • Assess the risk characteristics of a strategic asset allocation • Review relative value allocation decisions

The assumptions are not designed to inform short-term tactical allocation decisions. Our assumptions process is carefully calibrated and constructed to aid investors with strategic asset allocation or policy- level decisions over a 10- to 15-year investment horizon.

J.P. MORGAN ASSET MANAGEMENT 91 U.S. DOLLAR ASSUMPTIONS

Compound Return 2017 (%) Annualized Volatility Arithmetic Return 2018 (%)

Compound Return 2018 (%) Inflation

Inflation 2.25 2.26 1.25 2.25 1.00 U.S. Cash

U.S. Cash 2.00 2.00 0.50 2.00 0.08 1.00 U.S. Intermediate Treasuries

U.S. Intermediate Treasuries 3.00 3.07 3.75 2.25 -0.17 0.23 1.00 U.S. Long Treasuries

U.S. Long Treasuries 2.50 3.28 12.75 2.00 -0.23 0.04 0.78 1.00 TIPS

TIPS 2.75 2.90 5.50 3.50 0.07 0.07 0.65 0.51 1.00 U.S. Aggregate Bonds

U.S. Aggregate Bonds 3.25 3.32 3.75 3.00 -0.12 0.10 0.81 0.78 0.78 1.00 U.S. Short Duration Government/Credit

U.S. Short Duration Government/Credit 3.50 3.52 2.00 3.25 -0.09 0.41 0.76 0.41 0.67 0.74 1.00 U.S. Long Duration Government/Credit

U.S. Long Duration Government/Credit 3.25 3.68 9.50 3.25 -0.17 0.00 0.68 0.89 0.64 0.91 0.50 1.00 U.S. Inv Grade Corporate Bonds

U.S. Inv Grade Corporate Bonds 3.50 3.67 6.00 3.25 -0.12 -0.04 0.41 0.46 0.64 0.82 0.60 0.79 1.00 U.S. Long Corporate Bonds

U.S. Long Corporate Bonds 3.75 4.23 10.00 3.75 -0.16 -0.06 0.39 0.58 0.57 0.81 0.47 0.88 0.96 1.00 U.S. High Yield Bonds

U.S. High Yield Bonds 5.25 5.59 8.50 5.75 0.12 -0.10 -0.26 -0.27 0.31 0.19 0.15 0.12 0.57 0.47 1.00 U.S. Leveraged Loans

U.S. Leveraged Loans 5.00 5.28 7.75 5.00 0.32 -0.14 -0.50 -0.46 0.06 -0.07 -0.10 -0.10 0.33 0.24 0.80 1.00 World Government Bonds hedged FIXED INCOME

World Government Bonds hedged 2.50 2.54 3.00 1.75 -0.26 0.10 0.83 0.85 0.52 0.81 0.58 0.80 0.52 0.56 -0.19 -0.42 1.00 World Government Bonds

World Government Bonds 2.50 2.71 6.50 2.00 -0.01 0.13 0.66 0.48 0.66 0.71 0.68 0.59 0.58 0.54 0.17 -0.14 0.58 1.00 World ex-U.S. Government Bonds hedged

World ex-U.S. Government Bonds hedged 2.25 2.29 3.00 1.75 -0.27 0.07 0.71 0.75 0.43 0.72 0.50 0.72 0.50 0.54 -0.14 -0.36 0.97 0.51 1.00 World ex-U.S. Government Bonds

World ex-U.S. Government Bonds 2.25 2.56 8.00 2.00 0.02 0.11 0.56 0.37 0.61 0.62 0.62 0.50 0.55 0.50 0.24 -0.07 0.48 0.99 0.42 1.00 Emerging Markets Sovereign Debt

Emerging Markets Sovereign Debt 5.25 5.70 9.75 5.50 0.00 -0.02 0.22 0.16 0.58 0.60 0.46 0.50 0.76 0.68 0.72 0.40 0.28 0.55 0.29 0.57 1.00 Emerging Markets Local Currency Debt

Emerging Markets Local Currency Debt 6.25 6.94 12.25 6.50 0.12 0.12 0.15 0.02 0.46 0.41 0.41 0.31 0.56 0.48 0.61 0.30 0.14 0.59 0.13 0.63 0.82 1.00 Emerging Markets Corporate Bonds

Emerging Markets Corporate Bonds 5.25 5.57 8.25 5.50 0.05 -0.06 0.10 0.02 0.50 0.51 0.45 0.40 0.78 0.67 0.75 0.54 0.13 0.43 0.14 0.46 0.89 0.73 1.00 U.S. Muni 1-15 Yr Blend

U.S. Muni 1-15 Yr Blend 2.50 2.55 3.25 2.50 -0.06 0.03 0.46 0.47 0.52 0.66 0.45 0.57 0.58 0.53 0.26 0.10 0.52 0.43 0.49 0.38 0.51 0.27 0.39 1.00 U.S. Muni High Yield

U.S. Muni High Yield 4.50 4.73 7.00 4.25 0.21 -0.12 0.00 0.06 0.31 0.30 0.08 0.27 0.41 0.34 0.43 0.54 0.11 0.10 0.13 0.11 0.43 0.23 0.39 0.59 1.00 U.S. Large Cap

U.S. Large Cap 5.50 6.41 14.00 6.25 0.05 -0.06 -0.31 -0.35 0.06 0.00 -0.04 -0.03 0.27 0.23 0.69 0.55 -0.24 0.14 -0.18 0.21 0.52 0.59 0.55 -0.01 0.20 1.00 U.S. Mid Cap

U.S. Mid Cap 5.75 6.93 16.00 6.75 0.07 -0.08 -0.34 -0.35 0.07 -0.01 -0.05 -0.04 0.29 0.24 0.74 0.59 -0.26 0.10 -0.20 0.17 0.52 0.58 0.56 0.01 0.21 0.96 1.00 U.S. Small Cap

U.S. Small Cap 5.75 7.35 18.75 7.00 0.04 -0.08 -0.37 -0.38 -0.02 -0.10 -0.12 -0.11 0.18 0.15 0.65 0.51 -0.29 0.03 -0.23 0.10 0.42 0.51 0.46 -0.08 0.09 0.90 0.95 1.00 U.S. Large Cap Value

U.S. Large Cap Value 5.75 6.75 14.75 6.25 0.04 -0.07 -0.32 -0.35 0.02 -0.02 -0.05 -0.04 0.26 0.22 0.68 0.53 -0.23 0.13 -0.17 0.20 0.50 0.59 0.54 -0.03 0.17 0.98 0.95 0.91 1.00 U.S. Large Cap Growth

U.S. Large Cap Growth 5.25 6.19 14.25 6.25 0.07 -0.06 -0.32 -0.36 0.09 0.00 -0.04 -0.04 0.28 0.22 0.71 0.58 -0.26 0.12 -0.20 0.20 0.52 0.57 0.55 0.02 0.22 0.98 0.96 0.89 0.92 1.00 Euro area Large Cap

Euro area Large Cap 6.75 8.91 22.00 7.25 0.04 0.03 -0.21 -0.30 0.14 0.09 0.12 0.02 0.36 0.29 0.69 0.50 -0.19 0.32 -0.15 0.39 0.61 0.70 0.61 0.05 0.18 0.85 0.83 0.74 0.83 0.85 1.00 Japanese Equity

Japanese Equity 6.25 7.25 14.75 5.75 0.02 -0.09 -0.25 -0.21 0.12 0.08 0.05 0.09 0.38 0.34 0.60 0.46 -0.18 0.17 -0.14 0.23 0.49 0.57 0.55 0.01 0.15 0.68 0.67 0.60 0.66 0.68 0.70 1.00 Hong Kong Equity

Hong Kong Equity 6.50 8.39 20.50 7.50 -0.01 0.08 -0.18 -0.24 0.17 0.14 0.15 0.10 0.44 0.38 0.64 0.50 -0.15 0.23 -0.12 0.29 0.61 0.68 0.65 0.05 0.22 0.68 0.66 0.58 0.64 0.70 0.73 0.62 1.00 UK Large Cap

UK Large Cap 6.25 7.61 17.25 7.50 0.09 -0.01 -0.30 -0.36 0.12 0.05 0.05 0.00 0.38 0.31 0.73 0.63 -0.26 0.25 -0.20 0.33 0.60 0.66 0.64 0.04 0.29 0.86 0.82 0.72 0.83 0.86 0.90 0.70 0.78 1.00 EAFE Equity hedged EQUITIES

EAFE Equity hedged 6.25 7.12 13.75 6.50 0.00 -0.04 -0.40 -0.35 -0.03 -0.03 -0.08 -0.02 0.31 0.27 0.69 0.61 -0.26 -0.03 -0.18 0.03 0.52 0.54 0.57 -0.01 0.24 0.88 0.86 0.78 0.86 0.87 0.88 0.79 0.73 0.87 1.00 EAFE Equity

EAFE Equity 6.25 7.61 17.25 6.75 0.05 0.00 -0.24 -0.30 0.16 0.10 0.11 0.06 0.42 0.35 0.74 0.56 -0.20 0.31 -0.16 0.39 0.65 0.74 0.67 0.05 0.22 0.88 0.86 0.76 0.85 0.88 0.97 0.81 0.80 0.95 0.91 1.00 Emerging Markets Equity

Emerging Markets Equity 8.00 10.04 21.50 9.25 0.08 0.08 -0.19 -0.26 0.25 0.14 0.18 0.08 0.44 0.36 0.73 0.55 -0.17 0.32 -0.14 0.39 0.68 0.81 0.69 0.06 0.27 0.76 0.77 0.68 0.73 0.78 0.82 0.66 0.89 0.82 0.76 0.87 1.00 AC Asia ex-Japan Equity

AC Asia ex-Japan Equity 8.25 10.20 21.00 9.25 0.01 0.06 -0.18 -0.22 0.24 0.16 0.18 0.11 0.46 0.39 0.72 0.53 -0.13 0.30 -0.10 0.36 0.66 0.76 0.68 0.08 0.26 0.75 0.75 0.66 0.71 0.77 0.79 0.65 0.92 0.79 0.75 0.84 0.98 1.00 AC World Equity

AC World Equity 6.00 7.10 15.50 6.75 0.06 -0.02 -0.28 -0.34 0.14 0.06 0.06 0.02 0.38 0.31 0.76 0.59 -0.23 0.25 -0.18 0.33 0.63 0.72 0.66 0.02 0.23 0.95 0.93 0.85 0.93 0.95 0.94 0.77 0.80 0.94 0.91 0.97 0.89 0.87 1.00 Global Convertible hedged

Global Convertible hedged 5.00 5.45 9.75 - 0.03 -0.04 -0.33 -0.34 0.14 0.09 0.07 0.03 0.45 0.36 0.82 0.67 -0.23 0.12 -0.16 0.19 0.64 0.62 0.71 0.09 0.31 0.86 0.88 0.78 0.83 0.88 0.86 0.73 0.79 0.87 0.90 0.90 0.85 0.84 0.92 1.00 Global Credit Sensitive Convertible hedged

Global Credit Sensitive Convertible hedged 4.25 4.44 6.25 - 0.14 -0.07 -0.12 -0.19 -0.01 -0.01 -0.03 0.01 0.19 0.18 0.28 0.33 -0.10 0.08 -0.06 0.12 0.19 0.23 0.28 -0.01 0.22 0.37 0.34 0.29 0.37 0.35 0.38 0.31 0.35 0.44 0.41 0.41 0.31 0.30 0.40 0.39 1.00 Private Equity

Private Equity 7.25 9.21 21.00 8.00 0.06 -0.03 -0.25 -0.30 0.10 0.04 0.03 0.00 0.30 0.25 0.65 0.51 -0.20 0.19 -0.16 0.25 0.51 0.57 0.53 0.03 0.19 0.82 0.82 0.76 0.80 0.81 0.80 0.62 0.63 0.79 0.78 0.82 0.72 0.69 0.84 0.78 0.33 1.00 U.S. Direct Real Estate

U.S. Direct Real Estate 5.25 5.79 10.75 5.50 -0.01 -0.02 -0.04 -0.05 0.04 0.05 0.00 0.04 0.11 0.10 0.23 0.14 0.00 0.08 0.01 0.10 0.19 0.22 0.15 0.05 0.07 0.30 0.31 0.31 0.31 0.29 0.26 0.20 0.18 0.23 0.25 0.26 0.21 0.21 0.28 0.24 0.07 0.25 1.00 U.S. Value Added Real Estate

U.S. Value Added Real Estate 6.50 7.37 13.75 7.00 -0.01 -0.03 -0.05 -0.06 0.04 0.05 0.00 0.03 0.11 0.10 0.25 0.15 -0.01 0.09 0.01 0.10 0.20 0.23 0.16 0.05 0.08 0.33 0.33 0.33 0.34 0.31 0.27 0.21 0.19 0.25 0.27 0.28 0.23 0.22 0.30 0.26 0.07 0.27 0.95 1.00 European Direct Real Estate

European Direct Real Estate 5.75 6.72 14.50 6.25 -0.03 -0.01 -0.21 -0.19 -0.08 -0.08 -0.10 -0.08 0.05 0.03 0.28 0.25 -0.16 -0.08 -0.13 -0.06 0.17 0.18 0.17 -0.03 0.07 0.40 0.39 0.38 0.39 0.39 0.40 0.30 0.27 0.36 0.45 0.39 0.30 0.29 0.39 0.38 0.15 0.35 0.30 0.30 1.00 Asia Core Real Estate

Asia Core Real Estate 5.50 6.07 11.00 5.50 -0.01 0.05 -0.14 -0.15 -0.02 -0.06 -0.05 -0.06 0.04 0.03 0.23 0.16 -0.12 0.05 -0.10 0.08 0.14 0.25 0.10 -0.07 0.04 0.33 0.33 0.32 0.32 0.34 0.32 0.25 0.36 0.30 0.32 0.33 0.38 0.40 0.36 0.31 0.09 0.29 0.40 0.39 0.21 1.00 U.S. REITs

U.S. REITs 6.25 7.42 16.00 6.00 -0.05 -0.07 -0.02 0.00 0.21 0.27 0.09 0.25 0.44 0.42 0.62 0.36 0.10 0.29 0.13 0.32 0.56 0.59 0.47 0.21 0.24 0.74 0.76 0.74 0.76 0.70 0.64 0.52 0.48 0.59 0.63 0.66 0.56 0.56 0.71 0.62 0.18 0.63 0.40 0.43 0.32 0.29 1.00 Global Infrastructure

Global Infrastructure 6.25 6.89 11.75 6.25 0.10 -0.01 -0.10 -0.11 0.03 0.00 -0.02 -0.01 0.08 0.06 0.22 0.19 -0.08 0.05 -0.06 0.07 0.16 0.19 0.16 0.00 0.08 0.30 0.30 0.28 0.30 0.29 0.25 0.20 0.19 0.26 0.26 0.26 0.23 0.22 0.29 0.25 0.11 0.25 0.30 0.29 0.16 0.17 0.25 1.00 Infrastructure Debt

Infrastructure Debt 4.25 4.41 5.75 4.25 -0.05 -0.04 0.44 0.51 0.67 0.81 0.57 0.78 0.92 0.88 0.47 0.32 0.54 0.49 0.51 0.45 0.65 0.39 0.67 0.62 0.45 0.08 0.11 0.00 0.06 0.11 0.16 0.20 0.27 0.21 0.13 0.21 0.26 0.28 0.17 0.29 0.11 0.13 0.04 0.03 -0.04 -0.05 0.25 0.02 1.00 Diversified Hedge Funds

Diversified Hedge Funds 4.25 4.52 7.50 3.50 0.19 0.08 -0.41 -0.40 0.06 -0.10 -0.07 -0.10 0.22 0.16 0.60 0.65 -0.35 -0.04 -0.29 0.02 0.36 0.39 0.44 -0.06 0.37 0.66 0.69 0.58 0.61 0.71 0.65 0.59 0.61 0.73 0.74 0.71 0.68 0.66 0.73 0.79 0.43 0.61 0.13 0.14 0.30 0.27 0.34 0.20 0.15 1.00 Event Driven Hedge Funds

Event Driven Hedge Funds 4.75 5.13 9.00 4.75 0.23 -0.02 -0.46 -0.50 0.05 -0.11 -0.06 -0.15 0.27 0.19 0.77 0.76 -0.41 0.02 -0.35 0.11 0.47 0.52 0.58 -0.07 0.36 0.80 0.83 0.76 0.78 0.81 0.76 0.66 0.69 0.83 0.81 0.82 0.78 0.75 0.85 0.87 0.51 0.72 0.20 0.22 0.34 0.29 0.51 0.25 0.15 0.88 1.00 Long Bias Hedge Funds ALTERNATIVES

Long Bias Hedge Funds 4.75 5.27 10.50 4.50 0.14 -0.01 -0.41 -0.47 0.08 -0.08 -0.01 -0.12 0.30 0.21 0.75 0.67 -0.38 0.10 -0.32 0.19 0.53 0.62 0.61 -0.07 0.28 0.86 0.89 0.81 0.83 0.89 0.83 0.72 0.80 0.87 0.85 0.89 0.89 0.86 0.93 0.92 0.42 0.78 0.22 0.24 0.35 0.35 0.55 0.26 0.13 0.87 0.94 1.00 Relative Value Hedge Funds

Relative Value Hedge Funds 4.50 4.73 7.00 4.25 0.26 -0.02 -0.40 -0.42 0.16 0.01 0.03 -0.04 0.40 0.30 0.84 0.85 -0.34 0.01 -0.28 0.09 0.55 0.54 0.64 0.07 0.49 0.68 0.72 0.62 0.66 0.70 0.68 0.63 0.68 0.77 0.75 0.75 0.75 0.73 0.76 0.84 0.44 0.64 0.17 0.18 0.29 0.25 0.45 0.22 0.33 0.85 0.92 0.86 1.00 Macro Hedge Funds

Macro Hedge Funds 3.75 4.02 7.50 4.00 -0.10 0.18 0.12 0.10 0.26 0.20 0.27 0.19 0.27 0.24 0.17 0.06 0.18 0.31 0.20 0.31 0.23 0.31 0.20 0.11 0.15 0.21 0.21 0.12 0.17 0.25 0.28 0.23 0.25 0.31 0.23 0.31 0.35 0.32 0.30 0.31 0.14 0.23 0.04 0.04 0.08 0.13 0.14 0.05 0.22 0.54 0.30 0.37 0.29 1.00 Direct Lending

Direct Lending 7.00 7.46 10.00 6.75 -0.04 -0.07 0.29 0.33 0.61 0.74 0.53 0.70 0.97 0.91 0.68 0.47 0.40 0.49 0.39 0.48 0.77 0.57 0.81 0.56 0.46 0.32 0.35 0.24 0.30 0.34 0.39 0.41 0.48 0.43 0.36 0.45 0.49 0.51 0.42 0.52 0.21 0.34 0.12 0.12 0.07 0.06 0.45 0.10 0.93 0.30 0.37 0.38 0.52 0.24 1.00 Commodities

Commodities 3.75 5.06 16.75 3.75 0.26 0.08 -0.11 -0.27 0.27 0.05 0.18 -0.02 0.24 0.17 0.46 0.37 -0.24 0.35 -0.25 0.41 0.39 0.54 0.43 -0.09 0.12 0.44 0.46 0.38 0.44 0.45 0.48 0.34 0.49 0.60 0.34 0.54 0.61 0.52 0.55 0.47 0.28 0.44 0.10 0.10 0.09 0.18 0.26 0.15 0.13 0.52 0.57 0.61 0.55 0.41 0.27 1.00 Gold Gold 4.00 5.58 18.50 4.00 0.04 0.08 0.37 0.24 0.50 0.42 0.41 0.32 0.37 0.30 0.11 -0.08 0.25 0.53 0.19 0.52 0.35 0.40 0.33 0.25 0.11 -0.02 0.01 -0.04 -0.03 0.02 0.07 -0.02 0.18 0.11 -0.12 0.09 0.23 0.19 0.08 0.10 0.01 0.04 0.00 0.00 -0.11 0.00 0.06 0.00 0.36 0.10 0.06 0.14 0.09 0.41 0.33 0.46 1.00

92 LONG-TERM CAPITAL MARKET ASSUMPTIONS 2018 ESTIMATES AND CORRELATIONS

Compound Return 2017 (%) U.S. DOLLAR ASSUMPTIONS Annualized Volatility Note: All estimates on this page are in U.S. dollar terms. Given the complex risk-reward trade-offs involved, we advise clients to rely on judgment as well as quantitative optimization Arithmetic Return 2018 (%) approaches in setting strategic allocations to all of these asset classes and strategies. Please note that all information shown is based on qualitative analysis. Exclusive reliance on this information is not advised. This information is not intended as a recommendation to invest in any particular asset class or strategy or as a promise of future performance. Note that

Compound Return 2018 (%) Inflation these asset class and strategy assumptions are passive only–they do not consider the impact of active management. References to future returns are not promises or even estimates of

Inflation 2.25 2.26 1.25 2.25 1.00 U.S. Cash actual returns a client portfolio may achieve. Assumptions, opinions and estimates are provided for illustrative purposes only. They should not be relied upon as recommendations to

U.S. Cash 2.00 2.00 0.50 2.00 0.08 1.00 U.S. Intermediate Treasuries buy or sell securities. Forecasts of financial market trends that are based on current market conditions constitute our judgment and are subject to change without notice. We believe the information provided here is reliable, but do not warrant its accuracy or completeness. This material has been prepared for information purposes only and is not intended to provide,

U.S. Intermediate Treasuries 3.00 3.07 3.75 2.25 -0.17 0.23 1.00 U.S. Long Treasuries and should not be relied on for, accounting, legal or tax advice.

U.S. Long Treasuries 2.50 3.28 12.75 2.00 -0.23 0.04 0.78 1.00 TIPS

TIPS 2.75 2.90 5.50 3.50 0.07 0.07 0.65 0.51 1.00 U.S. Aggregate Bonds Source: J.P. Morgan Asset Management; data as of September 30, 2017, except hedge funds, private equity, real estate and infrastructure, as of June 30, 2017. Alternative asset classes

U.S. Aggregate Bonds 3.25 3.32 3.75 3.00 -0.12 0.10 0.81 0.78 0.78 1.00 U.S. Short Duration Government/Credit (including hedge funds, private equity, real estate, direct lending and infrastructure) are unlike other asset categories shown above in that there is no underlying investible index. The return estimates for these alternative asset classes and strategies are estimates of the industry average, net of manager fees. The dispersion of return among managers of these asset classes and U.S. Short Duration Government/Credit 3.50 3.52 2.00 3.25 -0.09 0.41 0.76 0.41 0.67 0.74 1.00 U.S. Long Duration Government/Credit strategies is typically significantly wider than that of traditional asset classes. Return estimates for direct real estate are unlevered. We have refined our REIT estimates for 2018 to reflect: U.S. Long Duration Government/Credit 3.25 3.68 9.50 3.25 -0.17 0.00 0.68 0.89 0.64 0.91 0.50 1.00 U.S. Inv Grade Corporate Bonds net REIT leverage; REIT vs. core composite sector differentials and a more granular treatment of currency differentials. Estimates shown for 2017 do not reflect these refinements. U.S.

U.S. Inv Grade Corporate Bonds 3.50 3.67 6.00 3.25 -0.12 -0.04 0.41 0.46 0.64 0.82 0.60 0.79 1.00 U.S. Long Corporate Bonds Intermediate Treasuries reflect the 1 to 10 years sector of the market instead of the 7 to 10 years sector in prior years. Diversified hedge funds and conservative hedge funds are multi- strategy hedge funds instead of funds-of-funds in prior years. Hong Kong Equity is represented by the MSCI Hong Kong index instead of the Hang Seng index in prior years. Correlation U.S. Long Corporate Bonds 3.75 4.23 10.00 3.75 -0.16 -0.06 0.39 0.58 0.57 0.81 0.47 0.88 0.96 1.00 U.S. High Yield Bonds figures shown are rounded to two significant figures, which may cause a loss of information. All returns are nominal. For reference index information, please visit our website.